Procedure and House Affairs Committee on Oct. 20th, 2026

[The Chair, Liberal, Chris Bittle]

I call this meeting to order.

Welcome to meeting number 42 of the House of Commons Standing Committee on Procedure and House Affairs. Pursuant to Standing Order 108(3), the committee is meeting to resume its study of artificial intelligence and Canada's democratic institutions.

Today's meeting is taking place in public in a hybrid format, pursuant to the Standing Orders. I don't think anyone is online. All members are here in person.

I would ask all in-person participants to consult the guidelines written on the cards on the table. These measures are in place to help prevent audio and feedback incidents and to protect the health and safety of all participants, including the interpreters. I would like to remind all witnesses that committee members, as you know, may ask questions in English and in French. If you need interpretation, please take a moment to familiarize yourself, but you are frequent flyers and you know what to do.

As a reminder, all comments should be addressed through the chair. If you wish to speak, raise your hand. The clerk and I will do our best to recognize you.

Before we begin, I'd like to take care of a quick administrative item. The clerk has circulated a draft budget. Is there any opposition or concern?

I see thumbs up and nods, so it is by consent that the draft budget is passed.

I would now like to welcome our witnesses. They are familiar faces: Eric Janse, Clerk of the House of Commons; Natalie Foster, clerk assistant and co-chair of the House of Commons AI working group; and Christopher Henry, director general of digital enablement and innovation and co-chair of the House of Commons AI working group.

I understand that Mr. Janse will deliver a five-minute opening statement. Then we will proceed to rounds of questions.

Mr. Janse, please proceed.

[Eric Janse, Clerk of the House of Commons]

Thank you very much.

Mr. Chair, members of the committee, thank you for inviting my colleagues and me to appear before you today regarding the opportunities, risks and implications arising from artificial intelligence for Canada's democratic institution.

As mentioned, I'm pleased to be joined by my colleagues, Natalie and Chris. Behind us on the backbench, we have representatives from digital services, legal, finance and procedure, so that we are ready to answer any and all of your questions.

AI is rapidly changing how information is created, analyzed and shared, and within the House's digital environment the use of AI-enabled tools is becoming increasingly prevalent. While there are possibilities for innovation across the parliamentary environment, we must also carefully consider how these new opportunities can be balanced with the protection of people, information, systems and processes, as well as with the key principles that are fundamental to parliamentary democracy.

A recent article in The Economist on AI-generated political speeches illustrates how quickly questions about authenticity, accountability and appropriate use are moving from the theoretical to the practical.

The House administration's role involves helping to establish a secure and trusted environment in which AI can be used responsibly. We have already been laying a stable foundation to prepare and protect the institution, members and employees. This includes clear guidance, appropriate guardrails, training and support, and carefully designed proofs of concept that allow us to learn before expanding.

In 2024, the House administration established the AI working group to provide immediate guidance on generative AI and to develop a long-term AI strategy. Its cross-organizational membership and co-leadership from the parliamentary business and technology sectors reflect the reality that addressing AI in our organization is a fundamental shift that requires expertise and collaboration from all sectors.

Following the June 2024 launch of AI guidelines for employees, the House administration consulted with the recognized parties and released AI guidelines for members and their staff the following year. Both sets of guidelines explain how to use generative AI safely and ethically, while the guidelines for members also address protecting their information and that of their constituents.

At the same time, the House administration has created an AI-dedicated space on Source, bringing together the guidelines as well as information on the security ratings of AI tools, writing effective prompts, acquiring AI tool licences, and other reference materials.

AI and cybersecurity content was also incorporated into the members' orientation program, as well as into awareness materials distributed as part of Cyber Security Awareness Month, a practice we intend to continue and expand this year.

For the more serious AI challenges, such as deepfakes or other security risks, the House has also developed clear guidance and coordinated support for members. Through a “no wrong door approach”, the House administration will coordinate assistance from all service areas on topics related to AI.

Finally, consultations with the recognized parties also identified the need for a secure generative AI tool. In response, the House is conducting a proof of concept with members selected by the whips to assess the viability of offering such a service. It allows for the exploration of practical use cases within an institutionally managed environment with stronger protections for our data. The proof of concept will help assess the technology’s value; understand its implications for information security, privacy and data governance; and identify the guardrails, training, support and resources required for broader implementation.

The House administration recognizes the importance of safeguarding the integrity, security and non-partisan nature of our parliamentary operations, while supporting informed and responsible AI adoption. Consistent with our mandate to anticipate members’ evolving needs and to innovate in support of your work and parliamentary democracy, we will continue to invest in the people, safeguards and tools required for the responsible use of AI at the House of Commons.

We welcome the committee’s guidance and support as we continue this work.

This concludes our opening remarks. We would be pleased to answer any questions you may have.

Thank you.

[The Chair, Liberal, Chris Bittle]

Thank you very much.

We'll turn to Mr. Cooper first for six minutes, please.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

Thank you, Mr. Chair.

Mr. Janse, you referenced the artificial intelligence strategy for 2025 to 2027. That strategy has not been made public—has it?

[Eric Janse]

I'll turn to Natalie, who's our strategy expert, to answer that one.

[Natalie Foster, Clerk Assistant and House of Commons AI Working Group Co-Chair, House of Commons]

At the moment, no. The strategy is available internally only. It has not been shared with the public.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

Could the strategy be provided to this committee?

[Natalie Foster]

Absolutely. We'd be happy to do that.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

Okay. Thank you for that.

You spoke about building a stable foundation for AI within Parliament, including clear guidance, appropriate guardrails and training and support. You spoke about some of the training for members. What about training for staff?

[Eric Janse]

Are you asking about the staff of members or the staff of the House administration?

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

It would be both the staff of members and of the House of Commons administration.

[Eric Janse]

I think it's all of the above.

Natalie, do you want to speak to that?

[Natalie Foster]

We've launched a training program for members and their staff, so both are invited to those. We have a monthly seminar that's available virtually. It's put on through our members' orientation program. That's open to members and their staff. The first version of that really focuses on the guidelines, guardrails and risks.

Then, through our proof of concept for Copilot, we've also established a training program specific to Copilot, its uses and use cases, and that is open to the participants of the proof of concept. That's meant to be a test to see if that's useful so that it could be expanded if ever we were to expand that as well.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

With respect to the training, what has been the uptake?

[Natalie Foster]

For the one that's open to all members, I believe we've had about 100 participants so far in that course. That was during the summer, so I could get some updated statistics for what may have been done in September. We have had a very high uptake.

In general, it's viewed quite favourably. We do feedback surveys afterwards as well.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

What about for staff?

[Natalie Foster]

That's a combined number of members and staff.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

Okay.

Regarding appropriate guardrails, can you elaborate on what those guardrails are?

[Natalie Foster]

I'll turn to my technical colleague to talk about some of the technical guardrails we have in place.

[Christopher Henry, Director General of Digital Enablement and Innovation, and House of Commons AI Working Group Co-Chair, House of Commons]

Thank you for the question.

We started with guidelines first to ensure the safe and responsible use of AI and to ensure that our people, including members and their staff, have the right information to be able to use these tools responsibly. That's been the premise of our human-centred focus for AI adoption here at the House of Commons. It's one of the pillars of the strategy, which we'll be sharing with the committee. Those guidelines were the first layer of defence we had to make sure those understandings were clear.

When it comes to guardrails, we look at those more from a technical controls perspective to see how we can further protect the environment for how AI is being used. Those start as the foundational pillars that we've had in our IT environment since the dawn of using IT for many different use cases pre-AI. Those include information security protections and visibility on all of our information so that we can have assurances that we can protect that information.

That's how we've started with the guardrails. It's simply from an administrative controls perspective to make sure we can have the right safeguards for members and their use of AI.

[Michael Cooper, Conservative, St. Albert—Sturgeon River, AB]

Thank you for that.

To what extent is the House administration using current AI tools?

[Eric Janse]

I'll start and then hand off to Natalie.

For instance, we are using AI in the parliamentary publications directorate for Hansard production. The first “take”, if you will, of Hansard is now produced in part using AI. That's just one of many other areas.

Natalie, do you want to expand on that?

[Natalie Foster]

With respect to Hansard and committee evidence, while the team does use that for a first pass, we employ the human-in-the-loop concept. There is always verification and validation, and we have our employees pivot very quickly to that higher-value work of editing and proofreading. While it's there as an assist, it's not meant to replace the expertise that our employees have.

The other use case that we're exploring is looking at transcription or closed captioning for private or in camera meetings that may be used for assistive devices for accessibility purposes. That's currently being explored, though it's really very much in pilot mode.

The third general use case might be around some coding practices, and I'll let Chris speak to that.

[Christopher Henry]

When it comes to the digital applications that support Parliament, since the creation of these applications, many of those that are used to provide support for proceedings, for example, have generally been of a high-code or human-coded aspect from a digital creation perspective. In the advent of exploring the capabilities of some of these generative AI tools, we have begun some light exploration of these tools insofar as, as Natalie mentioned, assisting our people with the ability to use those tools to create code for our products. Again, this is in a very small, limited testing base use case.

The final use case that I could add for House administrative use is general productivity. That's another area we're exploring as well—things such as summarizing meetings and planning our days—when it comes to understanding the capabilities of the technology.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

We'll now turn to Mr. Jeneroux for six minutes, please.

[Matt Jeneroux, Liberal, Edmonton Riverbend, AB]

Thank you, Mr. Chair.

Thank you to you three for being here today and also for bringing your staff along with you. There are a lot of familiar faces in the audience here. Thank you for providing the strategy to us. I think that will be helpful to know.

I want to pick up where you finished on using AI in Hansard production. I didn't know that was happening. It's probably a good sign that there hasn't been any sort of noticeable change in the process. I'm wondering when you started using it for the first run-through of Hansard. Have you seen any risks? Have you received general feedback, positive or negative, on that?

[Eric Janse]

Natalie, who's responsible for the Hansard department, can speak to that.

[Natalie Foster]

Once upon a time, I was.

It's been in place for about a year. It was launched in May of last year. It was a bit of a bumpy start in the sense that we wanted to make sure our teams understood the technology and were trained, so they weren't required to use it while we were figuring out those first few steps. We've maintained the same processes in terms of the review of the work, ensuring there are human editors at multiple levels reviewing the information before it goes to members through blues or is published for the public.

We've heard from the team that it's been a help ergonomically. People used to do the typing and sometimes sustained injuries from that repetitive movement. That has been removed because that technology does the first pass, and they can focus on the higher-value work of editing and proofing. In general, there's also been a request from our teams for more training for a better understanding of the technology and where it could go and the impacts that it could have on them and their teams.

[Matt Jeneroux, Liberal, Edmonton Riverbend, AB]

In terms of the level of accuracy, you haven't seen a difference.

[Natalie Foster]

There is a fairly high level of accuracy because it is trained on the historical records of Hansard. It's trained on members' names, members' constituency names and that sort of thing. There are still errors, and our technical colleagues are always involved in helping to further refine and train the model. This model is hosted on our premises. It's something that is within our infrastructure, so that data doesn't leave our systems.

[Matt Jeneroux, Liberal, Edmonton Riverbend, AB]

That's interesting. Are there any examples other than Hansard?

[Natalie Foster]

It applies to Hansard and to committee evidence as well.

[Matt Jeneroux, Liberal, Edmonton Riverbend, AB]

It's used for committee evidence too. That's interesting.

We've talked about the guardrails and the risks in terms of my previous colleague's question. Do you see a lot of general positives overall that members could take from using this, say, in our constituency offices? We obviously deal with a lot of private and sensitive information, so it's not just here in the House of Commons. A lot of that could be extended to our riding offices.

[Eric Janse]

It's a very good question. You're right that members have different roles, and the use of AI needs to be adapted or adjusted given the different roles.

To your point, constituency work is exactly that; it involves a lot of private information. I think, for that part, the proof of concept that's being undertaken allows the AI tool to access information that is confidential, but it doesn't allow the information out to the public. That, we hope, would be of particular benefit to members for all the casework that you do.

[Natalie Foster]

Our members' guidelines are very specific that you should not be putting personal or identifiable information into public tools. That's very important. There are other measures indicated in the guidelines that members should be following, but that's why we're working through the proof of concept for an internal tool where that data could have greater protections, where it could stay on our premises and where you would be assured that you could use that for productivity purposes.

With the question of AI, there are two sides. There's innovation, and there's risk. Members need to determine where they're comfortable and where they need our support to help make those determinations.

[Matt Jeneroux, Liberal, Edmonton Riverbend, AB]

I have two minutes, but this is probably my last question.

With reference to The Economist's article that you mentioned, Mr. Janse, are there flags that we should watch for, like giving speeches in the House of Commons using AI? We've seen what I would call some ridiculous examples of AI being used by parliamentarians around the world when giving speeches that include things like a disclaimer.

Have you explored that, perhaps, in some of the work that the committee has already done? Are there some suggestions that you would have for this committee as we undertake the study?

[Eric Janse]

Ultimately, members are still responsible for the speeches that they deliver, regardless of how they are drafted.

As per the guidelines, our advice would be to triple-check everything before you put it in your speech. You're right; we've all heard stories, even in court cases, of people providing information or precedents that, in fact, did not exist, etc. That is obviously something you want to avoid.

In terms of specific advice, Natalie, do you want to speak to that?

[Natalie Foster]

It's definitely something we're looking to explore further. We're currently looking at that, for example, within the committee context. We know that the public submits a large volume of correspondence and briefs, so we will have a proposal for the Liaison Committee to consider how they might want to look at disclosure for the public, in particular.

In terms of members, I really do think it falls upon your personal responsibility and accountability to determine what's appropriate for you, your caucus and the electorate in order to decide how you may want to engage, or not, with those disclosures. We'd be happy to support those discussions.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

Ms. Normandin, you have the floor for six minutes.

[Christine Normandin, Bloc, Saint-Jean, QC]

Thank you, Mr. Chair.

Witnesses, thank you for being here. I really appreciate it.

First, I would like to hear what you have to say about the pilot project you are conducting with Copilot. I understand that Copilot is part of the Office suite we're already working with, but I'd like to know what criteria you used to decide to take on Copilot as an artificial intelligence tool as part of your pilot project.

[Natalie Foster]

Thank you for the question. I can start by answering and then turn it over to my colleague Mr. Henry.

First, I just want to emphasize that this is a proof of concept, so this is just a start. We're running tests. We want to test use cases. We want to test the information security parameters. Since the Microsoft tool was already in our system, we found it easy to try it as the first testing tool. That doesn't mean it's the tool we will prioritize in the long term.

I will now turn it over to Mr. Henry so that he can tell you a bit about the security parameters.

[Christopher Henry]

Thank you.

Indeed, as Ms. Foster said, it's really because we're already using Microsoft applications, so it's something we can add to our applications. The general idea with the Copilot protections is that it's also part of the administration of our Microsoft platform. That means we can manage certain things with the Copilot parameters. That's one advantage of this proof of concept.

As Ms. Foster said, it's really just to make an assessment and learn how this technology is being used in our work here.

[Christine Normandin, Bloc, Saint-Jean, QC]

Outside of the project, members are free to use any platform. Have any risks been identified regarding the use of other platforms that could lead to a recommendation against the members using them?

[Natalie Foster]

I will talk about the strategic side, and then Mr. Henry can talk about the technological side.

We have a colour-coded list of tools on Source. Red means that the tool is not recommended. Black means that it's blocked. The other tools are in yellow. That means that they can be used, but that it's important to keep guidelines and responsibilities in mind, especially when it comes to the types of data that will be entered into these tools, because the data is not protected by our systems.

That's changing a bit. That's what we're trying to change with a proof of concept of a generative tool so that we can provide you with a service that is protected so that your data doesn't leave our system.

We have done an initial review of the tools on that list that are available to you internally to ensure that it's acceptable to enter data into them, but only public data.

Mr. Henry, do you want to add anything?

[Christopher Henry]

The problem is that each person has to know all the data they are entering into these tools and check whether it's acceptable under the guidelines. However, we sort of want to avoid having guidelines that force users to make a decision with each use. The general idea behind the proof of concept is to have an element of control that we can add to protect users and data.

[Christine Normandin, Bloc, Saint-Jean, QC]

In fact, I understand that the guidelines are recommendations. It isn't something that can be imposed on members. However, I would like to hear your thoughts on two aspects.

First, despite everything, is there any form of oversight of how members use various artificial intelligence tools?

[Natalie Foster]

I can start.

Yes, we have a cybersecurity team that regularly checks—every month, every few weeks—what tools are being used on our infrastructure. This doesn't target anyone; it's really general data. I think the team is exploring other options to dig deeper. However, we have monthly reports that list the tools being used on our system, and we can immediately look for solutions or potentially block tools that are unsuitable for use on our system.

I think we also want the opinion of the committee and members in general to know what should or should not be allowed and whether we should take a much stricter approach or keep a more permissive approach.

[Christine Normandin, Bloc, Saint-Jean, QC]

That's actually kind of my next question.

Has there been any discussion on imposing a regulatory framework rather than just having guidelines?

If so, would that look like, for example, guidelines but with an actual regulatory framework for certain more sensitive aspects? Is that part of the discussion?

[Eric Janse]

That's sort of part of the consultations with the parties, because it's certain that we would want to have agreement from all the parties before we impose anything.

Ms. Foster could talk about the consultations that have taken place so far.

[Natalie Foster]

We hold consultations twice, mainly with the whips and research offices of the parties, to find out your opinions, your requests and your needs. That's where our Copilot proof of concept comes from.

However, I think we have to study the issue carefully to determine how far we want to go. I know that there are other parliaments around the world where, if a tool hasn't been verified by the cybersecurity team, you aren't allowed to use it. It's about cybersecurity and data security. There are other parliaments that take a guidelines-based approach, a more open approach.

We're really trying to find out what you want. I think there is indeed a matter of security. Our cybersecurity teams will tell us when a situation is unacceptable, and we will then take action, that's for sure. However, apart from that, we're also looking for what you want in terms of support from the House of Commons administration.

[The Chair, Liberal, Chris Bittle]

We'll turn to Ms. DeRidder for five minutes, please.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you, Chair.

I have a couple of technical questions. With the development of AI, and considering data ownership with the mindset of cybersecurity, do you think Canada should be considering Canadian software as well as a Canadian cloud for storage of particular House business?

[Eric Janse]

I think the storage is part of the agreement.

[Christopher Henry]

Thank you for the question.

Yes. When it comes to data protections and looking at what may also be considered data sovereignty in this question and how we could examine that further, certainly our current state with our current supplier, from the Copilot and Microsoft perspective, is that we have agreements and data residency assurances through our contractual obligations with Microsoft from a data storage perspective. What that means for members and their staff is that your OneDrive, your SharePoint sites and any aspects of your work within the Microsoft suite enjoy the protections of Canadian data residency.

You mentioned Canadian software companies. Microsoft is not technically a Canadian-headquartered and -owned company. As such, one exploration we are considering as part of ongoing proofs of concept, to further examine how we can support members in ensuring that we're looking at all aspects of protection of their information and its use, is that we are exploring some other Canadian companies around the AI field. Those have not yet materialized an offering yet, but they are in the works from a planning perspective. We can examine further what it would look like if we were to offer more protections, not just residency but what I'll call independence and perhaps some aspects of control of the information.

These are areas we want to examine and explore further. The other side to examine as well involves the capabilities, because certainly with some of these might come some trade-offs when it comes to the features and what they might offer.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you very much for that. It's good to know that Canadian options are being explored, because we know that with data residency it's very important that it be Canadian-owned and Canadian-headquartered on Canadian soil.

You mentioned an agreement with Microsoft. Inside that agreement, does it protect us from the U.S. CLOUD Act?

[Christopher Henry]

Thank you for the question.

While I'm not an expert by any means on the legalities of the U.S. CLOUD Act, from a purely technical perspective, when it comes to the House of Commons' use of Microsoft as a technology and the data we use within the platform, so any of its applications, we have assurances in our contractual agreements for Microsoft to contact the Speaker and the law clerk and the various officials we've put in all of our terms and conditions of our contracts with our suppliers. That does offer a good level of protection.

As we examine aspects of risk of the CLOUD Act further, there are aspects that we are starting to notice. We see where there could be some legal aspects that could ask for access to the information, through court orders south of the border, where we may not be able to fully intervene, let's say, before those orders happen. There's always jurisprudence and aspects for us to examine those areas as they become known, but there are risks with the CLOUD Act around that.

To round out the answer, we don't know everything, certainly, about the aspects of the legalities, but part of our exploration in other solutions, such as a Canadian AI technology company, is that we could certainly examine how the aspects of the CLOUD Act would perhaps be minimized in our use of the technology through other vendors.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you for your transparency about the U.S. CLOUD Act.

Kitchener is Canada's innovation capital, I will proudly declare, so we have some options right in my hometown and they're working on solutions as we speak.

Am I out of time?

[The Chair, Liberal, Chris Bittle]

No, but be very quick.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Okay.

AI is changing very rapidly. Where do you see AI helping with workflows and remaining as a tool in your tool box to help increase efficiency? What are some of the key areas in the House where we can implement it?

[The Chair, Liberal, Chris Bittle]

Answer very quickly.

[Eric Janse]

I think Natalie mentioned a few earlier.

I don't know if you want to repeat...or if there are any others that we have further down that we envision looking at.

[Natalie Foster]

At the moment, I think general productivity is another area where we can see some gains. That's the summarization of public documents, potentially the summarization of meetings, and initial drafts of certain documents as well. However, I do think that maintaining the accountability and the human loop is very important to anything that we would go forward with.

[The Chair, Liberal, Chris Bittle]

Thank you very much.

Mr. Caley, you have five minutes, please.

[Braeden Caley, Liberal, North Vancouver—Capilano, BC]

Thank you very much Mr. Chair.

Thank you to the witnesses today. It's good to see several of you again.

To switch gears just a tiny bit, I thought it could be useful to hear a little bit about the accountability side of things and a bit more about the public's interaction with the House of Commons.

Before asking that, I'll compliment you on the work that's been done on this so far. In general, the House of Commons tech infrastructure.... As a new MP, it's been neat to see how it's more creative and seemingly more embracing of new tools than a lot of other government entities in the country. Accountability-wise, I think there's a lot of potential here, and I may have time to get to things we should worry about as well.

Think about a certain democratization of the capability to analyze, of analysis, that might have been in the domain of think tanks or only well-funded news bureaus in the past. Now, with public tools, a member of the public would be able to compile their own questions and answers about the long arc of a political party's positions, statements or votes on matters.

From what you've seen, what excites you, either in what's been possible so far or in other international examples, about any potential for improvement to the accountability that could come—or ease of accessibility to the House by the public—because of this new technology and how fast it's accelerating?

[Eric Janse]

That's a very good question. We could probably spend half an hour on it.

Very quickly.... I think the AI tools give access, or easier access, to all the information that's out there. That, I think, is already a beneficial advantage for Canadians in general, especially those who are following parliamentary proceedings closely. I know as well—because there are information management experts who help committees and members and the like—that the Library of Parliament is exploring how AI can be used to help it do its work for you better and more quickly.

Do you have something to add, Natalie?

[Natalie Foster]

Obviously, you've alluded to transparency and accountability being fundamental to the democratic process. I do think that as the House of Commons, as an institution, it is our duty to ensure that the data we create, disseminate and build here.... You can think about data around votes, proceedings and what's said, the spoken word, in the House and in committees. We do feel that we have a responsibility to ensure that information is made available to the public through open data as much as possible so that they can leverage it to be a better-informed citizenry and to be able to use that information in the way they see fit.

I see that as a possible future for us: to continue to go down that vein so that our information is more easily accessible to the public in the variety of ways they'd like to access it.

I don't know if you want to add anything to that.

[Christopher Henry]

Just to build on that, I will say that we currently are getting a lot of requests on our public websites, which are currently the main panes of glass for organizations, as you mentioned, to get our data. As Natalie mentioned, open parliaments, open data concepts, and standards around that are not new. However, more than ever, in the AI age, there's a need to democratize that data so that companies and stakeholders alike can repurpose and use that information with fewer barriers than they have today.

[Braeden Caley, Liberal, North Vancouver—Capilano, BC]

Are there particular barriers right now that you're sifting through as those requests increase?

[Christopher Henry]

Certainly, yes. It's a balance between.... When there are, let's say, AI crawlers, if you will, accessing our websites, that can impede the performance of and access to those websites for the purpose they are designed for. It's a catch-22 where we're trying to be open and available for all stakeholder needs while ensuring that at least the one window of glass on our information that we currently share is available at the same time. Certainly, something in our road maps to look at, as Natalie mentioned, is having more access to the data itself instead of to a presentation layer of the data to be harvested.

[Braeden Caley, Liberal, North Vancouver—Capilano, BC]

Both are important now.

This next question is a bit more technical and picks up on questions from a few colleagues.

Is the pilot of Copilot the same instance or same initiative as within the public service, or are these developing in parallel to each other? Are there any best practices from both being developed in parallel—if I'm reading the head shaking correctly—that are being taken so far?

[Christopher Henry]

I'll start, and then if Natalie would like to add anything....

Through the Parliament of Canada Act and our institutional independence, yes, I can assure you that the Copilot pilot instance is within the House of Commons Microsoft agreement and our contractual separation from the other departments and agencies.

That being said, we collaborate extensively with our colleagues across the departmental areas, and legislative areas around the world, to ensure that we learn from each other how best to administer these tools and configure them in a way that is mindful of the risks and, of course, benefits for our clients.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

We'll now go to Madame Normandin.

Before I do, cognizant of the time, I think what I will do is give four minutes, six minutes and six minutes to the next three questioners. That will take us to....

I'm getting a look. Is that fine? Okay. I think that puts us in a good spot to get to switching the panel.

Madame Normandin, you have four minutes, please.

[Christine Normandin, Bloc, Saint-Jean, QC]

Thank you very much.

Ms. Foster, I'd like to come back to how you mentioned that zero non-public information can be entered into artificial intelligence tools. I would like to hear your comments on what that means for House employees. For example, I understand that there's a formal ban on using artificial intelligence to draft bills, translate bills or transcribe an in camera committee meeting, for example. Is that correct? If so, how can we ensure that it's properly enforced?

[Natalie Foster]

We have guidelines for our employees as well. There are similarities with the guidelines for members. No, people can't enter private information into public tools.

That said, when it comes to writing the Hansard, as well as committee evidence, we're talking about a protected tool in our infrastructure. In that case, we have permission to enter testimony or the Hansard into those systems, even in camera, because it's protected; it's within our infrastructure.

Translation always goes through our partner, the translation bureau. Artificial intelligence tools are not used to translate the Hansard, evidence, official publications or documents you receive in committee, for example. We always work with that partner.

When it comes to bills, we do have a team of law clerks and lawyers who support you in that work. They're the ones working on the bills, and no information is entered into public tools.

[Christine Normandin, Bloc, Saint-Jean, QC]

Since we're hearing from the interpreters and the translation bureau in a little while, I would like to hear your comments on one thing: Are there certain aspects of the House's work for which artificial intelligence is used?

One thing that comes to mind, for example, is the closed captioning used on ParlVU. Is artificial intelligence already being used? If it isn't—or even if it is—is there any intention to use it down the line? How would that work?

[Natalie Foster]

I can start by answering, and Mr. Henry can add to that.

The closed captioning in the House and in committee is provided by our partner, the translation bureau. As a result, the use of artificial intelligence tools to support this work may be a question for the bureau.

As I mentioned, the only aspect that might make us consider this is accessibility. For in camera meetings, private meetings, artificial intelligence could be used to provide closed captioning. It wouldn't be to replace the expertise of our partners; rather, it would be an accessibility measure for the people taking part in those meetings.

[Christine Normandin, Bloc, Saint-Jean, QC]

I have one last question, which is a bit more technical. The guidelines you provided recommend not using a House address to create an external account on an artificial intelligence tool. However, we use a House address for Copilot. I would like to hear your comments on the reasons behind that. Why is it acceptable for Copilot but not for other tools, for example?

[Natalie Foster]

I would ask Mr. Henry to answer that question.

[Christopher Henry]

This policy existed before the age of artificial intelligence. When it comes to using a cloud-based tool, the rule is always to not use parliamentary accounts. The general idea is that our parliamentary identity and the data associated with that identity are not used by software outside Parliament. The difference between Copilot and the Copilot pilot is really our control of the tool. By managing the data in this tool, we have more opportunities to guarantee and see how the data is used and managed by our teams here, to help you.

[Christine Normandin, Bloc, Saint-Jean, QC]

Thank you.

[The Chair, Liberal, Chris Bittle]

Thank you.

Mrs. Block, you have six minutes, please.

[Kelly Block, Conservative, Carlton Trail—Eagle Creek, SK]

Thank you very much, Mr. Chair.

Thank you to all of you for being here today.

I want to circle back to some of the questions that my colleague, Ms. DeRidder, was asking.

Mr. Henry, you said something that I found very interesting when we were talking about Canadian sovereignty and Canadian-owned, and keeping our data within our borders. You talked about there being trade-offs when we are looking to ensure that we own our data, that it's being housed here and that it's sovereign. Would you be willing to share with us what some of those trade-offs might be?

[Christopher Henry]

Yes. Thank you for the question. I'm happy to answer it.

For trade-offs, in this aspect, I meant capabilities, features and usage of the tools insofar as they offer value in the work. Obviously, risk and opportunity or value are things we constantly balance when we look at the risk of data sovereignty, which is data independence and an aspect of control. With regard to some elements, the companies that are, let's just say, offering these advantages may not have all of the capabilities that some of the frontier companies offer in terms of advanced capability.

That's what I was inferring a little bit around when it comes to trade-offs. It is from an example I've been hearing coming out of the European Parliament because there have been some discussions there regarding sovereignty and the spectrum of sovereignty. It's not a binary switch. It's a gradient, a continuum. If stakeholders are seeking greater degrees of sovereignty.... There's been an observation in the European Parliament. They've seen trade-offs insofar as capabilities they had enjoyed previously with some of tools that may have been less sovereign.

Again, this remains early days in our exploration. It's something we're keen on investigating further; hence, we're doing a lot of these proofs of concept.

[Kelly Block, Conservative, Carlton Trail—Eagle Creek, SK]

Thank you.

A question I have is this: How do you create digital sovereignty and protect your data when a handful of multinationals are creating the technologies and building the service?

[Christopher Henry]

I'll jump right into that. Essentially, there are companies within Canada that we have been in contact with, and with which we are exploring further, as I mentioned, plans to begin, in parallel, offering proofs of concept so that we can observe the trade-offs together while also being able to observe the advantages from a sovereignty perspective.

Again, sovereignty isn't simply data residency or Canadian-owned companies. It's really the degree to which Parliament can maintain some independence on its information and its data, so can we observe that this is actually something we can achieve?

[Kelly Block, Conservative, Carlton Trail—Eagle Creek, SK]

As you mentioned, the European Union.... Surely there are countries around the world grappling with the very same issues that we are as we look to implement AI on a much broader basis. Are we having dialogues with other like-minded democracies that have started implementing AI technology? Have we started them? Are we part of them? Are we leading in any way?

[Christopher Henry]

In this aspect, especially when it comes to risks of AI in general and how we can work together, the Five Eyes organization recently published—a few months ago—a call to action around the risks that it is observing around AI and its threat landscape. That's why this committee is very timely in terms of our coming to participate and share some information so that we can understand and protect better as we continue our use of AI responsibly.

[Natalie Foster]

I might also add, if I can, that we do have partnerships with other legislatures. We are speaking with other legislatures more from a parliamentary business perspective. What are they doing? What are they not doing? What are they avoiding? We all share, very freely amongst each other, guidelines, things we've learned and things we'd like to avoid, so we're building that community as well.

[Kelly Block, Conservative, Carlton Trail—Eagle Creek, SK]

That raises another interesting question for me. Are we leading here in our own country, with the provinces and the legislative assemblies here in Canada as well?

[Eric Janse]

We actually have excellent relationships with all the provincial and territorial legislatures. There is an association of the clerks. There's an administrative group. Also, the sergeants-at-arms of the different legislatures have a network.

Through those networks, there's been quite a bit of discussion on AI, because of course it's such an important topic, very much so. There's a lot of discussion so that nobody is reinventing the wheel. We hope to and intend to be leading that charge given the resources we have.

[Kelly Block, Conservative, Carlton Trail—Eagle Creek, SK]

The last question for me would be this: Are there other parliaments and other governments around the world looking to or coming to Canada for support in doing the very same thing that we are?

[Natalie Foster]

In general, I would say that other Westminster parliaments have been frequently in contact with us. We've been sharing our guidelines and even aspects of our strategy with them, so that we can build and learn from each other. In particular, I would say that the U.K., Australia and New Zealand are key partners, beyond our provincial and territorial partners.

We also have connections with other parliaments throughout the world. Some are forging ahead and building all sorts of AI tools. I would say that most are more in the posture that we're in of trying to be cautious and trying to determine the right balance between risk and innovation.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

For our last questioner, we'll go to Madame Mingarelli for six minutes, please.

[Giovanna Mingarelli, Liberal, Prescott—Russell—Cumberland, ON]

Thank you, Mr. Chair.

Thank you to all the witnesses for being here today.

My first question will be for you, Mr. Janse.

As we've discussed today, of course, Parliament is built not only on rules and information, but also on judgment, debate and human relationships. In order for AI to be trustworthy to the people it is serving, it has to be reliable and verifiable.

How do we ensure the proper verification of the data being submitted? Is it human or AI or both? Then, how do we account for hallucinations in the data when, basically, AI is making stuff up and presenting it as real. We believe it's real, but it's not. Sometimes it's very difficult to determine.

[Eric Janse]

Maybe I'll start. My colleagues can jump in.

It's a very good question.

Natalie mentioned committee submissions from witnesses. Sometimes a committee can get inundated with all sorts of submissions if people feel strongly about a certain study or a certain bill that the committee is looking at. Nowadays, you could get 1,000 to 10,000 submissions from one person in his basement, so we are looking both from a technical perspective but also from a guidelines perspective at how to try to control that or at least flag it for members.

One option would be to insist that as part of the submission process, they have to tick off a box saying that it was in part AI-generated but always verified by a human before we hit “send” to the committee. Maybe our technical team could be asked to take a deep dive into all these submissions, if you notice that there's a bit of a spike in submissions, to see to what degree we can determine, as best we can, whether or not they've been AI-generated or whether they're all genuine. There are opportunities for us to assist in that manner.

Natalie, I don't know if you have anything to add.

[Natalie Foster]

Yes, maybe I could just add on there.

On the responsibility front, the approach that's been taken in the U.K. and in Australia is to have witnesses attest that the information is theirs that they're submitting, and they're responsible for that information. That's the way those parliaments have chosen to address that question.

We will have a proposal for the Liaison Committee, but to reassure this committee, we have not seen levels of abuse or submissions that are concerning to us. Our excellent teams of clerks and analysts are always there to support the committee should there be concerns.

[Giovanna Mingarelli, Liberal, Prescott—Russell—Cumberland, ON]

That's great. Thank you.

That somewhat answers my next question: How do you ensure that only verified content is being presented as responses?

We would be pursuing the U.K. and Australian models for that.

[Natalie Foster]

I think that's something we'd like to put before the Liaison Committee for their consideration. I think the question for members is, what is it that you need to know in order to ensure the submissions you're reviewing are appropriate? It would be up to the members to tell us what you need. Then we'd be happy to support.

Our online submission form for briefs also has a CAPTCHA, which allows you to check “I am a human” when you're submitting that information. It would prevent abuse from bots, for example. We also have process protections. Those same protections apply to our e-petition process, which is also one of our other major interactions with the public in terms of parliamentary tools.

[Giovanna Mingarelli, Liberal, Prescott—Russell—Cumberland, ON]

Thank you.

Therefore, CAPTCHA content, the mechanism for verifying the individual, can be helpful, but it does have its own obstacles because it does not verify who the human is. You could have a person who is a spammer, who's sitting there in their basement for 10 hours one day submitting whatever they feel like for any number of reasons and clicking the CAPTCHA's “I am a human”.

Has any thought gone into the verification of the individual submitting the content so that if there are abusive claims or content that is submitted, or questionable content—foreign interference type of content—that individual can be traced to its originator by tracking IP addresses, these kinds of things, while maintaining privacy? How do we account for that?

[Natalie Foster]

We have been discussing with our technical colleagues some of these protections, so maybe I'll let Chris elaborate on that. I think it would be quite difficult to detect whether something is AI-generated, but there are safeguards and things we could put in place to determine if the same individual, let's say, were submitting multiple briefs from a similar IP address or within a short period of time. We would be able to exercise some of those technical controls.

I don't know whether you want to add to that.

[Christopher Henry]

Those are very good questions.

The basic answer is that we don't have all of the capabilities and sophistication to AI-detect all content being submitted to the House. Therefore, the same aspect that we would impose on stakeholders checking a box, we also on our side as we're receiving briefs and information that come from the public.... While we could equip our people with tools to look at and, perhaps, assist with detection, ultimately, our human judgment is applied to make decisions based on what is coming in. Then we apply countermeasures when necessary, such as contacting the individuals or the person in their basement and stating that we may not accept those materials as such.

As to going so far as to track their location and others, we'll stay away from that one. However, we're certainly doing our utmost on the “human side” protection by our administrative staff, to make sure that we do detection there.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

I'd like to thank the witnesses, and I would like to thank the army of potential witnesses behind them.

We will suspend for a few minutes for the next panel.

[The Chair, Liberal, Chris Bittle]

Welcome back, everyone.

I would like to welcome our second panel of witnesses. From the International Association of Conference Interpreters, Canada region, we have Alionka Skup, president, and Marina Mistral, conference interpreter and member of the AIIC science hub. From the Westminster Foundation for Democracy, we have Franklin De Vrieze, head of practice and accountability.

We'll go first to our witnesses who are present here. Then we'll go to our witness online.

You have five minutes, please.

[Alionka Skup, President, International Association of Conference Interpreters - Canada Region]

Good afternoon.

On behalf of the International Association of Conference Interpreters - Canada Region, or AIIC-Canada, I would like to thank you for this invitation. I would also like to acknowledge that we are meeting today on the sovereign and unceded territories of the Algonquin Anishinabe people.

AIIC-Canada covers the Canadian region of AIIC and represents 3,000 members across the country. Many interpreters work for the highest authorities in the world.

Today, we're here to talk to you about artificial intelligence and interpretation. Marina Mistral is our expert here to tell you about this topic.

[Marina Mistral, Conference Interpreter and Member of the AIIC Science Hub, International Association of Conference Interpreters - Canada Region]

Good afternoon, Mr. Chair and committee members. Thank you for the invitation to participate in this study.

My name is Marina Mistral, and I'm an accredited parliamentary interpreter. I would like to say hello to my excellent colleagues in the booth, with whom I have shared the booth in the past. My background as an engineer and my interest in technology have led me to sit on the new scientific committee of the International Association of Conference Interpreters, which is looking at the issue of artificial intelligence in interpretation, among other things.

Artificial intelligence is undoubtedly a very powerful tool for interpreters, but we believe that entrusting it with the mike carries risks of misinformation, the erosion of bilingualism in Canada, the breakdown of democratic discourse and the increased vulnerability of our institutions.

First, the use of machine interpretation carries technical risks that other experts have already spoken to you about in this study. Outages, or even system shutdowns for maintenance, could jeopardize the continuity of Canadians' language rights. A breach could make everything that happens here behind closed doors public. Administrative friction could prevent terminology updates from being frequent enough to keep up with current events.

The greatest risk with AI interpretation lies in the fact that artificial intelligence is not, in fact, intelligent. It is a statistical pattern-matching black box that may sound human but doesn't actually think. Therefore, it cannot be trusted to make the kinds of decisions that interpreters make every day to communicate effectively. I'll give you some examples.

From a sense-making standpoint, we can distinguish information from noise and unimportant words. We can make sense of unfinished sentences and subtext. When we hear a homophone, we know which meaning is intended. We can follow someone's reasoning and use intonation to facilitate comprehension. We can understand and convey tone, irony, humour and cultural references.

On the practical side, we can adjust our pace or reformulate when interpreting from English to French, which requires more words to say the same thing. We understand regional accents, dialects and proper nouns. Unlike AI, we can notice when we make mistakes, and we correct ourselves. We will flag technical problems when they arise and let you know when your microphone is off. All these choices make communication reliable and effective and require real reasoning, which AI is incapable of by nature and unlikely to be able to do in the foreseeable future.

We believe AI should never be given the job of interpreter, because nobody deserves AI-grade interpreting. Interpretation that cannot be trusted comes with serious repercussions and reputational risks. With the risk of public-facing information being distorted in unpredictable ways, the dialogue that supports our democracy could be weakened.

When MPs cannot trust what they hear, they cannot presume to be heard. How can they, then, represent their constituents? If francophone constituents, as the linguistic minority, are forced to choose between leaving the conversation or having it in English, it means they cannot participate meaningfully, and certainly not equally. It's a violation of their rights.

To build Canadians' trust in the use of artificial intelligence and set an example for the rest of the world, Canada must promote the responsible use of these tools, meaning only in contexts where their output is verifiable. With that in mind, if we compare translation and interpretation, it's entirely possible to revise a machine translation to make it a high-quality final document. That is a possible use.

With interpretation, it would be physically impossible to listen to two audio feeds at the same time and make real-time corrections, in addition to all the human resources required.

Even though artificial intelligence is not suited to the dynamic process of interpretation, it can serve as a valuable ally to interpreters. At the preparation stage, these tools now enable us to translate and summarize texts, quickly consult a variety of sources and generate glossaries on subjects that we're sometimes called upon to master in under an hour. Artificial intelligence tools could also help quickly create reliable schedules and provide each committee with the best-suited team, while respecting the many constraints that are currently causing the coordination team a lot of headaches.

We believe that Canada should adopt these best practices for using artificial intelligence and provide employees and suppliers with training and support regarding the responsible use of these kinds of tools, while continuing to listen to them as the technology evolves.

We would now be delighted to answer any questions you may have.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

Before we get to questions, we'll turn to Mr. De Vrieze for five minutes, please.

[Franklin De Vrieze, Head of Practice Accountability, Westminster Foundation for Democracy]

Good afternoon. I'm speaking from London.

I'm Franklin De Vrieze.

Thank you very much for the opportunity to address you today. It's a true pleasure joining you.

I speak on behalf of the Westminster Foundation for Democracy. In terms of background, the WFD is the U.K.'s public body for international democracy support. We are linked to the FCDO, the Foreign, Commonwealth and Development Office, as an arm's-length body with the mandate to provide international democracy support. We have around 30 country offices around the globe, where we work with national parliaments and subnational parliaments in those countries on a wide variety of themes and topics relevant to democratization.

With regard to this topic we are discussing today, there are three themes that we are focusing on. These are the digitalization of parliaments, the introduction of AI in parliaments and the question of cybersecurity. The three themes are, of course, linked to each other, but they are still very distinct.

I have just a couple of words at the start of my contribution. From our observation, working with so many parliaments, what do we see regarding why parliaments consider using AI-enabled tools? It has to do, of course, with the expectation of enhanced efficiency, increased transparency, better public engagement, accessibility, etc., but there are also expectations that go much further. For instance, there's predictive analysis or expectations around enhanced legislative drafting. Of course, these are the issues that are much more sensitive.

Another thing we have observed is what we call the pacing problem, which is the growing gap between the speed of technological change and the speed of change at political and decision-making levels. We see this pacing problem increasing, and it's harder and harder for political institutions to keep up with the speed of change at the technological level.

Specifically for parliaments, we see this pacing problem at three levels. We see an external pacing problem between parliaments and industry and society, but often—and I think this is important here—there's a pacing problem between the different branches of government. In quite a number of countries, we see parliaments sometimes struggling to keep up with how governments or executive agencies are introducing AI in their functioning. The effect of that is executive agencies, public services and government departments in many countries sitting on a mountain of data, which, of course, parliaments do not have at their disposal, although parliaments are supposed to exercise oversight and some control functions.

I think the issue of executive dominance is an increasing problem in many countries around the world with regard to or on the occasion of the introduction of AI.

We see AI, as you've just discussed, introduced in various ways in parliaments. I think the most obvious one is at the administrative level in terms of transcription, translation, document classification and summaries. The more advanced and the more delicate one, as I mentioned, is the introduction of AI tools with regard to the analysis of amendments, checks of legal consistency and information retrieval.

There are a couple of interesting examples of this. For instance, the Italian Parliament has piloted AI-supported compliance checking of amendments. The French National Assembly, for instance, has created tools to model fiscal and social impacts of legislation and also to compare draft bills with existing legislation.

There are plenty of opportunities with regard to the introduction of AI in parliaments, but there are also a number of risks, and they have already been discussed in your previous conversations. As a short introduction, this is what I wanted to share with you today.

Thank you once again. I'm looking forward to the conversation today.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

We'll now turn to questions. We'll start with Mr. Calkins for six minutes, please.

[Blaine Calkins, Conservative, Ponoka—Didsbury, AB]

Thank you, Chair.

I'm going to start my questions with you, Mr. De Vrieze. Thank you for joining us. I don't know what hour it is where you are, but I'm sure you'd rather be somewhere else. Thank you for being with us here at the committee today.

I have a question for you because this is something that's on my mind a lot when it comes to the use of artificial intelligence, and you brought it up. I don't think we've had a good conversation about this yet.

Here in Canada, it takes anywhere from three to nine months for a regular piece of legislation to go through its process. This is from the time it's introduced at first reading. It then goes for healthy, robust debate at second reading, and if it passes, it goes to committee. The committee will then sit and take anywhere up to two to three months to go through deliberations on the bill, make its amendments, send the bill back to the House of Commons—in the Westminster system, of course—and have more debate at third reading just in time to send it all to the Senate, where it goes through the exact same process again. If there are no changes being made, it gets royal assent from the Governor General on behalf of His Majesty the King.

That all takes months. How fast does AI evolve in three to nine months? That's how long it takes us to pass one bill.

[Franklin De Vrieze]

That is a very accurate question, but it's also a question to which there are very few easy answers to give, because the question doesn't apply only to the functioning of Parliament but to the functioning of public services in general.

You described a process of six to nine months. AI can function in the legislative process best when it's very targeted. That means when it's looking at, for instance, text consistency, cross-referencing and checking legislation or amendments with existing legislation. I think it can also contribute by providing a basis on the ex ante impact assessment of legislation based on the available sources.

Specifically on the issue of impact assessment, the experience of the French is that human verification and cross-verification are very important. That's why I say AI doesn't provide an impact analysis. It provides a draft or a basis to conduct an impact assessment.

The easier elements to use AI in the legislative process are around terminology alignment, maybe translation support and also, I would say, identifying normative gaps. Of course, no normative gaps are intended, so we're looking for unintended normative gaps and sometimes unintended consequences or, specifically, consequences for specific groups in a society that might be overlooked or might not be accurately assessed. These are issues around predictive analysis where AI tools are being piloted and these things are explored.

I don't think we are yet at the stage where we have a reliable system and a reliable set of guidelines with regard to this. I think we are still very much at the level of piloting. Piloting is good and experimentation is good, but experimentation and piloting are not the same as transformation. We are not yet at the stage whereby the legislative process as such will be transformed. We are still looking at ways it can be improved with the help and assistance of these various AI tools.

I'm sure you have heard this comparison. I'll conclude my answer on this question with this comparison. AI tools are a bit like an intern. We all love interns. Interns do very good work if they are well trained, if they have had the time to familiarize themselves with various issues and if there is good oversight and supervision of the role of interns. It's the same with regard to the role of AI tools. In that sense, I would exercise caution with regard to the use of AI in the legislative process, but on various other issues, with regard to communication and administrative issues and citizen outreach, I think it can provide useful contributions.

[Blaine Calkins, Conservative, Ponoka—Didsbury, AB]

I would agree with that. I do believe, though, notwithstanding that every parliament, every country, tries to pass technology-neutral laws when it comes to governance, that the rapid growth and change in artificial intelligence is something that our Parliament, and many parliaments, will have to wrestle with. As far as being nimble enough, our parliaments and our democracies have to be nimble enough to respond to sudden and unexpected changes in terms of where artificial intelligence might be taking our societies.

Notwithstanding that, I am also concerned about information sharing with the public. We heard from witnesses at a previous meeting of this committee that in a recent sub-sovereign provincial election in Canada, I think about 48% of the information that was gathered and researched for that particular election was attributed to a website that was generated by artificial intelligence from a political hobbyist.

What should our Parliament be doing to make sure that the information that's verified, factual and true is actually the information that artificial intelligence is using when it comes to communicating what's actually happening with our democracy at the federal level in Canada?

[The Chair, Liberal, Chris Bittle]

That was very long, but we will need a very brief answer. I'm sorry.

[Franklin De Vrieze]

Thank you very much. It's an important question.

I would say that the role of Parliament as an institution—not individual MPs, but the role of Parliament as an institution—with regard to outreach to citizens can be helped with some specific AI tools, such as an AI-powered chatbot at the Parliament website. The key issue, of course, is that the information included in this chatbot to respond to citizens' inquiries needs to be based on Parliament information only, not on something that the chatbot will search on the internet, because then there is a risk for hallucinations. Chatbots in reaching out to citizens can be useful if the framework for the source of information is clearly defined.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

We'll turn to Madame Brière for six minutes, please.

[Élisabeth Brière, Liberal, Sherbrooke, QC]

Thank you, Mr. Chair.

Thank you to all our witnesses.

Thank you very much for your remarks.

Ms. Mistral, you mentioned the fact that artificial intelligence is a very powerful tool for interpreters. I imagine you use it for certain tasks. Can you give us some concrete examples of uses that don't compromise the quality of your service? For example, I'm thinking about looking up words in a specific sector or taking notes.

[Marina Mistral]

Here in Parliament, we use artificial intelligence a lot, because we're in a context where we often have to prepare very quickly for committees and meetings. Sometimes we only have a few minutes.

It's now possible to use confidential—hence, paid—versions of artificial intelligence. Those enable us to ask terminology questions. For example, I was working on a committee—I believe it was the Standing Committee on Fisheries and Oceans—for which I had to very quickly put together a termbase on a type of fish and on crab fishery. That means it was very specific to a given situation.

It's possible to ask questions and put together glossaries, but it's very important to then verify those glossaries using official Government of Canada resources and to consult studies previously done by the government to ensure that we're actually using Canadian terminology. That's really important. Acronyms can often be a problem, so we need to verify them. We need to verify regional ways of saying things, because artificial intelligence still tends to base its French on European French content. That's one example, then.

It's also possible to use transcription in some cases, but, again, it's important to really verify what we're reading.

My colleague can add examples of uses, but I would ultimately say that we can use artificial intelligence for terminology extraction. This can be very useful if we have two texts that have already been translated by the translation services here. It's possible to extract very frequent terms or technical terms.

That's how we normally use AI. We also sometimes use it to summarize a topic. If we provide material or documents, in particular, we can ask for a summary that, again, we can verify with our human intelligence.

[Élisabeth Brière, Liberal, Sherbrooke, QC]

I'll lighten the mood a bit. In my previous job, I once received a translation generated by artificial intelligence. The translation of “legal seal” came out as “phoque légal”.

I'm struck by another point. You spoke about the decline of bilingualism. I would like you to elaborate a bit on this issue. Do you think that the quality of interpretation might vary between French and English? We also know that indigenous languages are spoken in the forums. I would like to hear your comments on these two aspects.

[Marina Mistral]

Good question. I would like to provide an example, because I didn't have time to do so in my opening remarks. I can send you the sources.

The World Health Organization, a United Nations agency, has already looked into this issue. It assessed the quality of artificial intelligence in interpretation with the help of a team of professors from Geneva. Only 1 out of 90 speeches received a passing grade. All the speeches posed some kind of reputational risk, because the artificial intelligence often failed to identify proper nouns, country names or people's names. It also made gender‑related errors and did some truly objectionable things.

French, as the source language, received the lowest score at 36% . This is noteworthy because the interpretations into English carried the highest number of reputational risks. This was also compounded by a delay of over 30 seconds in the interpretation. The figures and technical information posed real problems.

I think that, at this point, the quality of this technology is nowhere near sufficient to support Canadian democracy, especially given that francophones are a minority here. The same goes for indigenous people, of course. We also know that Canada has a thriving francophonie that stretches from the Atlantic to Manitoba, including Quebec, of course. Each community has its own regional accents and terminology.

You can already see this in Moncton. You may have heard that the city's conference interpreter has retired. The city has had to turn to the largest provider of artificial intelligence‑based interpretation services. You can listen to the interpretation yourself. It's a total disaster. The city will hire an interpreter as soon as possible. In addition to a delay of up to a minute, the machine is basically unable to understand anything spoken in French or English. It often makes hilarious mistakes, which certainly pose a reputational risk.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

Ms. Normandin, the floor is yours for six minutes.

[Christine Normandin, Bloc, Saint-Jean, QC]

Thank you.

Ms. Skup and Ms. Mistral, I would like to continue discussing two of the points that just came up.

First, as you said, interpreters are in short supply, especially in Moncton. Would it send the wrong message if parliaments started using artificial intelligence to recruit and train interpreters, who remain crucial? Are wrong messages already being sent? In your opinion, would this send another wrong message that could give us many headaches in the future?

[Alionka Skup]

I think that young people who have an interest in languages are already concerned. They don't want to pursue a career in this field given all the talk about artificial intelligence replacing interpreters and translators. This risk is already a reality.

We're already seeing some institutions—even international institutions such as the Organization of American States—conducting trials with artificial intelligence. The results have proven disastrous, particularly for French, once again. It seems that artificial intelligence is being trained less frequently with French.

This certainly sends a frightening message to people who are interested in getting into the field. Rather than being a solution—as we're being led to believe—it further exacerbates the current shortage.

[Christine Normandin, Bloc, Saint-Jean, QC]

Let's talk about delays in circumstances where, like here, speaking time is crucial. Sometimes we have only two and a half minutes to ask a question. If the interpretation takes an extra minute, we might miss the beginning of a sentence. When the conversation switches to another language, we might miss the rest of it. We could end up missing half the remarks as a result of poor interpretation timing.

I just want to hear you say it again. What's the added value of an interpreter who can adjust the length of sentences given that the timing must be right in order for us to catch everything?

[Marina Mistral]

The delay stems from the nature of machine interpretation. For now, and likely in the short to medium term, the process begins with a transcription. This step alone is prone to errors, especially for languages other than English. The machine then performs the translation and reads the new text. This whole process causes a delay, depends on the technology and remains vulnerable to attacks.

Obviously, the training program for interpreters here in Parliament includes the most demanding exam in the country to ensure that we can keep up with the conversation. Our delay is no more than five seconds, and that's really only in the worst‑case scenario. We're often able to summarize the remarks and rephrase them. We've received training. Our experience helps us convey the meaning of what people are saying, rather than simply translating words. This is really important. You're thinking individuals who convey ideas. We're also thinking individuals who convey your ideas, because we understand you.

[Christine Normandin, Bloc, Saint-Jean, QC]

You brought up various aspects. Interpreters can pick up on expressions, intonation and cultural differences. I would like to hear your thoughts on this. Is the non‑verbal aspect being addressed? When interpreters look at us, can they also pick up on some of the information through non‑verbal cues, which would be absolutely impossible for artificial intelligence to detect?

[Alionka Skup]

It's certainly one of the things that we use. Like everyone else, when we speak with someone, we respond to non‑verbal cues. We also understand certain aspects as interpreters and as human beings who can do this. Clearly, artificial intelligence can't access this information at all. It truly takes a human's understanding of other humans to pick up on non‑verbal cues. That's absolutely certain.

[Christine Normandin, Bloc, Saint-Jean, QC]

You spoke about disastrous experiences in certain places, particularly in Moncton. To your knowledge, have any other parliaments around the world dared to use artificial intelligence for interpretation?

[Marina Mistral]

There have been internal tests. For now, the AIIC Science Hub's research shows that no one has yet dared to use publicly available artificial intelligence‑powered interpretation. So far, the trials have been limited to internal communications that have not been made public or broadcast. As I said, the current test results have been absolutely disastrous across the board. We don't really know when this issue will be resolved. Security is also an issue, of course. At this point, the security protocols aren't really strong enough for institutions to fully commit and become that dependent on a provider or technology for important and confidential matters.

[The Chair, Liberal, Chris Bittle]

Mr. Moore, you have five minutes, please.

[Rob Moore, Conservative, Fundy Royal, NB]

Thank you, Mr. Chair.

Thank you to the witnesses for their important testimony today.

I marvel at our interpreters and how, in real time, they're able to give a translation of what we're saying in the House, whether it's during question period and some of the very heated exchanges that happen there or whether it's in a setting like this at committee.

You touched on this a bit. I don't know if you're in a position to say, but I want to get your thoughts on, basically, the loss of human capacity. You mentioned some examples where AI was tried and was a dismal failure. I think you already mentioned the chilling effect it has on someone considering this for a career path, but I could foresee very quickly a loss of that human capacity. We're worried about that in other areas, but should there be a widespread adoption of AI for this particular service....

It was interesting. I was talking to someone yesterday, and they mentioned whether something was a good translation or a bad translation. Once we lose that capacity, who's to even say whether something, from a human perspective, is a good or a bad translation if we were to turn all of that over to AI? I want to get your thoughts on that. It's a very specialized field.

Could you speak to how people get into that field and to the sustainability of having people like you able to do the job you do?

[Marina Mistral]

I think everybody in our profession is worried right now, even the people who are already employed and my colleagues at the UN institutions, about the decisions that are being made by management. We see it in the translation services, where professional translators are now having to spend their days reading what sometimes amounts to AI slop and very bad, convoluted translations. Everybody in our profession is absolutely worried about the decisions that management everywhere is going to make.

One of our two Canadian programs, the one at Glendon campus at York University, is right now experiencing extreme difficulties. It has been paused for a couple of semesters. We're having issues everywhere with recruitment.

There is absolutely no doubt that the climate we see right now.... There is also even some discourse that we hear about some contexts deserving good interpretation and some contexts deserving less good interpretation—not professional, accredited interpreters but people who are in the market and who have not undergone the training we undergo or do not have experience and are now being employed by the federal government.

Of course, it is a worry, and it is very difficult now to convince somebody to choose this profession that we absolutely love and hope to be doing until we retire, basically.

[Rob Moore, Conservative, Fundy Royal, NB]

Thank you.

Do you have a sense of where Canada stands globally amongst peer countries in adoption? If one is a 10 and another is a one, with one being no adoption whatsoever, where would you say that Canada stands right now?

[Marina Mistral]

I would say that, in Canada, we have the ICAO, which is an agency of the UN. It follows the UN policies for AI adoption, so it has its own rhythm. I wouldn't say this is necessarily representative of Canada, but it has started doing some big groundwork on translation only. There is no talk about AI interpretation yet, even if some members would like to investigate the question.

Here we know that just about everywhere in the world, and in Canada as well, we are doing some automatic translation that in most cases has to be verified by a human. That is really important because we don't know what comes out of AI because it's such a black box. I think just about everybody has started doing that.

I would say that the bigger institutions in Europe have more power or more budget to try to really do some studies and research, as the WHO has done and I have mentioned. It was a very large study that was conducted last year about AI interpretation in a real context.

I would say that we are behind in the sense that we do not have this very large body with many languages that the UN and the European institutions have.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

Ms. Vandenbeld, you have five minutes, please.

[Anita Vandenbeld, Liberal, Ottawa West—Nepean, ON]

Thank you very much.

I'd like to direct my questions to Mr. De Vrieze.

Franklin, it's good to see you again. Most of your experience—for more than 25 years—has been with parliaments, and then AI is secondary. Parliaments are a very unique kind of place. We're not the same as other institutions that might be adopting AI. What would you see as a difference for the adoption of AI by a parliament?

You mentioned that we're not interpreting legislation; we're drafting legislation. We have privilege. We have in camera testimony that needs to be taken into consideration, but we are also responsible and accountable to the public. How do you see the differences for AI in a parliamentary setting?

[Franklin De Vrieze]

Thank you very much. It's good to see you, Ms. Vandenbeld. Thank you for the question.

Indeed, parliaments are very unique institutions. I think the introduction of AI is at different levels. It is about the introduction of AI in the functioning of a parliament, but it's also about the role of a parliament in overseeing the policies and the legislation with regard to the introduction of AI to society as a whole and to the public sector as such. I think that is a main difference.

The second thing I would say is that the introduction of AI in a parliament is very much a change process. It's not a tech project. We should not consider the introduction of AI as something that has to do with IT only or procurement. It is really about an institutional change process. It means that in deploying tools without addressing the underlying issues of data quality, governance, skills and also organizational culture, the results and the gains will be marginal and could even undermine trust.

I would say that effective change in a parliament with regard to AI is very much about establishing cross-parliamentary governance bodies, establishing a parliamentary consensus on the introduction of AI—and consensus among the different parties but also consensus between the political level and the administrative level—and also establishing what I would call continuous learning rather than one-off training.

Indeed, as was mentioned before, the development of AI is going very fast, so it is really about institutional learning. We have been developing quite a number of tools around that, including some AI guidelines for parliaments—initially drafted in English and now translated into 10 languages. That was also quite a unique experience, discussing it in so many different parliamentary contexts. Some parliaments are still at the level of digitalization, and others are very much on the AI trajectory. I think that, in Canada, you're very much ahead of the game on that.

There are also very serious risks, which we haven't really grasped yet. There is the risk very briefly mentioned with regard to cybersecurity. I mean, when we speak about cybersecurity, it's like a time bomb. Cybersecurity, for Parliament, is like a time bomb that, with AI, is put on high alert. Before, a cyber-attack was something that a highly skilled IT person would be able to do after months or weeks of preparation. Now, cyber-attacks are happening on a daily basis against parliamentary infrastructure. We've seen that in a number of countries, a number of parliaments: Estonia, the German Bundestag, the U.K. Parliament, the Norwegian parliament and, of course, the Ukrainian parliament. This is a very worrying trend.

The head of cybersecurity in one European parliament told me that the parliamentary infrastructure is being attacked approximately 25,000 times a day, and then you have three or four ICT officers who are trying to protect the infrastructure. It's not just the email addresses or the phone numbers of the members of Parliament—the attacks go farther. The attacks go through the legislative database. When legislation is published, it's not being printed on paper: Is that not right? That's how it was published before. Publication of legislation, after months of debate and after proclamation, is then being published online. Now, what does it mean being published online? It's on a portal. If AI systems are being used to change the text of the legislation as it is published online, then, of course, AI is playing its most destructive role in the legislative process.

You, as members of Parliament, can debate policies and legislation and vote, but is the text you see on the screen really what you have voted for? I think that the question of cyber-protection of the legislation, as published online, is a critical issue, which at the moment is really underestimated. That's why the issue of the use of AI by malign actors looking for the weakest points in a state infrastructure and usually the offices of prime ministers—

[The Chair, Liberal, Chris Bittle]

I have to cut you off there. You have my apologies. We're a little bit over time on that round.

Ms. Normandin, you have the floor for two and a half minutes.

[Christine Normandin, Bloc, Saint-Jean, QC]

Thank you.

I'll once again turn to the representatives of the International Association of Conference Interpreters.

You spoke about the fact that some large institutions have enough money to research the effectiveness of artificial intelligence. However, I would like to hear your views on the reasons behind a parliament's potential decision to use artificial intelligence for interpretation. From what I'm hearing, it seems that the only good reason to do so would be to save a few pennies. It certainly wouldn’t be to increase efficiency.

Other uses of artificial intelligence show that it helps us become more productive, find and keep track of information, summarize information, and so on.

Aside from the financial benefit, is there any advantage to using artificial intelligence in interpretation?

[Alionka Skup]

No. The only benefit, really, is saving money. We're told that the goal is to address a shortage. However, it isn't a real shortage. The more artificial intelligence becomes widespread, the greater the shortage of interpreters will be. It's really just a matter of saving a few pennies. People will settle for “good enough” and will provide interpretation services only because they need to.

I would also add that this really undermines the French language. You should know that 90% of the interpretation and translation in Parliament is into French. This is really just a way to pinch pennies at the expense of the equality of Canada's official languages and at the expense of francophones in particular.

[Christine Normandin, Bloc, Saint-Jean, QC]

How much does this heighten fears that the French language will suffer, given thePrime Minister's plans to cut the translation bureau's budget by 25% over five years, even if these cuts don't directly affect interpretation?

What message does this send, this desire to save money at the expense of the French language and the quality of the interpretation?

[Alionka Skup]

It seems a bit hypocritical to call bilingualism a priority while making these budget cuts. The translation bureau provides both interpretation and translation services. It's a 25% reduction for both these aspects.

Furthermore, this affects not only services provided to Parliament, but also services provided to the various departments supported by the translation bureau. The impact extends beyond parliamentarians to include the people of this country, who have the right to access high‑quality interpretation.

[Marina Mistral]

I would also like to add that the savings in question here aren't necessarily real. These savings are being dangled in front of us. However, we know perfectly well that, when we end up with a single provider for an essential service, we may see cost overruns that actually far exceed the cost of interpretation here today.

We've seen this happen here often. As soon as a supplier has a captive market of this nature and a major client such as the Government of Canada, I highly doubt that the supplier will be willing to help the Government of Canada save money.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

We'll turn to Ms. DeRidder, please, for five minutes.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you, Mr. Chair.

My first questions are for you, ladies.

I agree that interpretation should remain human. That being said, can you identify any key areas where you think AI could be implemented to actually help you as a tool to increase efficiency in House business?

[Marina Mistral]

I think coordination is probably the main piece for House business. We know how difficult it is to coordinate the schedules here because of the many last-minute changes. We interpreters receive schedules very late at night. Usually the coordination team is making a tremendous effort to fit this puzzle together. It can be very difficult. You need a specific language combination. You need to prepare the teams. You need to give them a certain amount of time between assignments and so on.

I think AI is excellent for doing this. For example, the product of organizing an interpreting schedule for the House could be easily verified and the errors could be flagged. It could not only save hours; it would also make things run smoothly for everybody. It would make life much better for interpreters as well, because then they could maybe count on more stable schedules. They could plan their lives and basically have better jobs. That's especially for staffers.

Of course, the tools we use to prepare for assignments are very precious. If we could get access to these tools as providers and staff, we could also be trained on how to use them responsibly. That means confidentiality. It's important to educate all the users of AI that when we are dealing with confidential information, it's important to use the appropriate version of the products. It's important to educate people on how to use it in such a way that we diminish the risk of the message being lost. I think this would be very helpful.

Almost all of us use these tools already. They're very useful.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you very much.

You mentioned that internal testing is being conducted with AI interpretation, and the results are disastrous. Do you happen to have access to those reports?

[Marina Mistral]

Absolutely. I will send you all the links after this committee meeting is done. I have quite a few that science has produced.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you very much.

My last question is for Mr. De Vrieze.

You mentioned the speed at which technology is advancing and the lack of policy to keep pace with the advancement, as well as that the adoption within our own parliamentary systems is lacking. Where do you believe are two key areas in which AI can be adopted into our House business to improve public service?

[Franklin De Vrieze]

Thank you very much for this question. I think in the areas of citizen communication, citizen outreach, responding to citizens' queries, the participation of citizens and reviewing the input and the submissions of citizens, the responsible use of AI can be done.

Second, I would say that comparisons between legislation, and comparisons between amendments and existing legislation, are probably where AI could also be used responsibly. I'm not speaking about drafting legislation but rather about comparing legislation and amendments within the country as well as across different jurisdictions.

I think that's somewhere AI could be used responsibly and relatively safely.

[Kelly DeRidder, Conservative, Kitchener Centre, ON]

Thank you very much.

That's all I have.

[The Chair, Liberal, Chris Bittle]

Thank you.

We'll now turn to Ms. Kayabaga for five minutes, please.

[Arielle Kayabaga, Liberal, London West, ON]

Thank you, Chair.

I'll turn to the interpreters first.

You have said a great deal. To some extent, I agree with you on interpretation. I think that you said earlier that no progress could be made, even in the future. I would like you to explain why you're so certain that we'll never reach the point where artificial intelligence can work, despite the current trends.

That said, do you have any recommendations for our committee? What specific measures, in order of priority, should we consider?

Our colleague asked you to provide information on the results of the tests you have already carried out. We would also like to see this information.

I would like to hear your thoughts on the future advancement of artificial intelligence in interpretation and then hear your recommendations for this committee.

[Marina Mistral]

I'm well aware that artificial intelligence is advancing rapidly. That's why I didn't provide any examples of funny things that artificial intelligence is saying right now. The situation is sure to improve. My examples were more about human reasoning. The machine can never do this because it's a statistical black box that relies on statistical models and simply doesn't think. My examples concerned comprehension or rephrasing, for instance.

There's also the matter of realizing when we make a mistake. When we realize that the context doesn't quite match the output, we can make a correction. The statistical black box will never be able to do this type of thing. It's designed to generate a probable result. I think that certain situations in simultaneous interpretation absolutely require human judgment and that the machine will never be able to make that judgment.

In terms of priorities, I think that Canada should take the lead, as it has been doing. When it comes to interpretation, Canada is already seen as a role model for bilingualism. We have an incredible collection of bilingual documents. We truly set an example for other countries around the world. I think that we should continue to integrate artificial intelligence responsibly and to demonstrate how artificial intelligence tools can be used for preparation and for terminology extraction. It's really about helping interpreters get on board with this technology and giving them training. On the administrative side, it’s also about using artificial intelligence to manage interpreters' work and to help allocate resources.

I'm sure that artificial intelligence can become more powerful. A 60‑second delay could be reduced to 30 seconds or 20 seconds. However, can we really expect this delay to decrease sufficiently in the near future?

Furthermore, can we really accept that the members of Parliament, who use a rich variety of expressions, have their remarks interpreted by completely impersonal machines? As we know, the lack of intonation significantly increases the cognitive load, making it difficult to listen to a machine‑generated interpretation for more than a few minutes. Is it acceptable for your words to be interpreted by an automated process? I don’t think so.

[Arielle Kayabaga, Liberal, London West, ON]

I agree with you. There are different accents in Canada. I'm a francophone from southwestern Ontario, so my accent is different. Even our vocabulary differs from the words used in Quebec or Acadia, for example. I agree with you and I'm really glad that you're here today.

[The Chair, Liberal, Chris Bittle]

Thank you so much.

I'd like to thank all of our witnesses for appearing today.

This meeting stands adjourned.