Skip to Content

Facebook Is Enlisting Human Experts and AI to Fight Terrorism

Technology alone won’t defeat the social network’s problems with extremist content.
Heightened security was put in place on the streets of New York City after London's recent terror attacks.

Following recent terror attacks, politicians criticized social networks for providing safe spaces for extremism. Now Facebook has announced that it’s developing AI and employing a team of 150 experts in order to become "a hostile place for terrorists."

Mark Zuckerberg has explained that, in his utopian vision of the future, many of the problems that his social network faces—among them violent content, child abuse, fake news, and extremism—will be eased by the development of artificial intelligence. Powerful algorithms, the theory goes, could sniff out offensive content and rogue users to shut the problems down.

But to date his vision has yet to materialize. Instead, fake news is being fact-checked by third parties and thousands of people are employed to sift through potentially offensive content. That's all for the good, but it's not the automated solution that Zuck promises.

Now, though, Facebook has described how it’s at least making use of artificial intelligence to attenuate the successes of extremist recruitment and propaganda. In an announcement, Monika Bickert, the company’s director of global policy management, and Brian Fishman, its counterterrorism policy manager, explain that the social network’s use of AI against terrorism is “fairly recent,” but argue that it’s already having an effect.

It even gives some concrete examples of what it’s doing:

  • using image recognition to identify previously removed content so that it never makes a reappearance
  • analyzing text to understand language associated with terrorism and spot propaganda
  • measuring social ties between regular and rogue users to identify recruitment activities
  • and monitoring how repeat offenders create new accounts to shut them down sooner

The big question is, of course, whether any of this will work. Other experiments using AI to solve knotty social issues online—such as Google’s bid to stamp out hate speech—haven’t come to much. A large factor is the fact that humans still often find ways to outsmart even the cleverest of AIs: in Google's case, by subverting language so that AI doesn’t identify the true meaning of a statement, for instance.

Facebook knows that to be a problem. Indeed, it says that “this work is never finished because it is adversarial, and the terrorists are continuously evolving their methods, too.”

That may be why it’s admitting that technology alone is not enough. The company has also built a team of 150 counterterrorism experts—including former academics, prosecutors, and law enforcement agents—to help contextualize content and spot real-world problems that might require intervention. It also confesses to the fact that it’s not an island, so it will continue to work with other social networks and governments to crack the problem.

By combining human insight with AI grunt work, Facebook is of course maximizing its odds of success in stamping out a problem that has plagued it for years. But only time will tell if together they’re smart enough to outwit the extremists.

(Read more: Facebook, “If Only AI Could Save Us from Ourselves,”  “Fighting ISIS Online,” “Theresa May Wants to End 'Safe Spaces' for Terrorists on the Internet. What Does That Even Mean?”)

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.