One challenge—and regulators have experienced this in other jurisdictions—is that we use AI to do so many things on a spectrum, from lightly modifying to significantly changing or completely fabricating material. Determining where something is on the spectrum is difficult. Quite often, it's clear that synthetic content is being fabricated, so it's not problematic.
In addition, we do not currently have great and reliable detection systems in order to identify these things, so it could be very difficult for a regulator to figure out when someone is transgressing a requirement to reveal synthetic content.
For these reasons, I think it could very well be that this is something for which we want to further develop the capacity to ensure that we can identify manipulated or fabricated content. However, I think that the way the content is being used and what it is aiming to achieve are more critical, whether that is making false claims about how to vote, harassing candidates during an election and so forth. That's why I think the emphasis should be there.
I doubt that there are even—