This is part seven of the Facebook files, an ongoing series from the journal.

We're examining internal Facebook documents to understand how the company operates and

affects the world.

If you haven't listened to the previous episodes, be sure to check them out during your feed.

Members of the Subcommittee, thank you for the opportunity to testify.

In 2020, Mark Zuckerberg testified in front of Congress.

During the hearing, he touted Facebook's use of artificial intelligence to police the

platform.

He said Facebook's AI was doing a good job of taking down bad content that breaks the

company's rules like hate speech.

In terms of fighting against hate speech, we built AI systems and as I mentioned, have

tens of thousands of people working on safety and security with the goal of getting the

stuff down before people even see it.

And Zuckerberg put a number on how effective those AI systems were.

Right now, we're able to proactively identify 89% of the hate speech that we take down before

I think it's even seen by other people.

So I want to do better than 89%.

I'd like to get that to 99%.

This is a really important statistic because it shows that the company is taking the bull

by the horns and really getting deep and preventing the perpetuation of hate speech across the

platform.

That's our colleague, Deepa C. Theraman.

...