Sure, I'd be happy to do that.
I can obviously only speak for how we think about it at Facebook. I don't want to make representations for other companies. Again, it is clear that we were much too slow to identify this new kind of threat back in the U.S. presidential election. When we did turn our mind to it—and I was getting at it a bit with an answer to Mr. Saini—we were trying to look at automated signals to understand, from a political standpoint, foreign interference, and how we could recognize that on the platform. It turns out that it's people setting up fake accounts on Facebook and spreading misinformation.
Now, as you know, on Facebook we have an authentic identity policy. Overwhelmingly, the two-plus billion people on Facebook actually behave a certain way because they're real people. They will do things in their personal time that we expect normal people to do. The fake accounts actually behave very differently. With AI we are able to identify, and we're getting increasingly better at identifying, these fake accounts and taking them down proactively. When you look at the subsequent elections after the U.S. presidential election, when you look at the French election and the German election, and most recently at the Italian election, you will see that we were able to identify tens of thousands of fake accounts and take them down proactively. I'm pleased to say that, although, again, our work is never done—