Instagram Starts Using Artificial Intelligence to Moderate Comments. Is Facebook Up Next?


Instagram started to automatically block offensive comments Thursday, using artificial intelligence to go beyond simple keyword filters. The use of this technology is also a test case for Facebook as it is looking to improve its own moderation and filtering.

The Facebook-owned photo sharing service officially announced the launch of a new comment filter Thursday morning. “Many of you have told us that toxic comments discourage you from enjoying Instagram and expressing yourself freely,” wrote Instagram CEO and co-founder Kevin Systrom in a blog post. “To help, we’ve developed a filter that will block certain offensive comments on posts and in live video.”

Instagram also announced a new spam filter, which it had quietly been testing over the last couple of months. Filtering abusive comments will for now only be available in English, but spam is being detected if it’s written in Spanish, Portuguese, Arabic, French, German, Russian, Japanese, or Chinese as well. Comment filters are enabled by default, but can be turned off by each user.

Both filters are powered by machine learning, which means that the technology used to filter comments has been trained with a test set of data, and is looking at not just keywords but also contexts and relationships. An f-word between friends may have a completely different meaning than a slur hurled at an outsider, and song lyrics can include a lot of offensive language without actually offending anyone.

Instagram’s comment and spam filters are based on DeepText, an artificial intelligence effort developed in-house at Facebook, as Wired reported Thursday.

That’s notable in part because Facebook itself has yet to officially commit to AI as a means to moderate content and comments. Executives previously said that it may take some time before AI can play a role in moderation, and Facebook has responded to recent controversies related to inappropriate content with the hire of thousands of additional human moderators.

However, more recently, the company seems to have changed its tune on AI a bit. “There’s an area of real debate about how much we want AI filtering posts on Facebook,” wrote CEO Mark Zuckerberg in a blog post earlier this month. “It’s a debate we have all the time and won’t be decided for years to come. But in the case of terrorism, I think there’s a strong argument that AI can help keep our community safe and so we have a responsibility to pursue it.”

Does this mean that Facebook may also eventually use AI to moderate comments the way Instagram is now? A Facebook spokesperson stressed that both platforms are unique in a statement emailed to Variety:

“Facebook and Instagram are different platforms with different user experiences — from the follow model to how comments are used. Although we share the same goal of creating safe communities, we are going to have different approaches. Instagram’s new tools are a great first step that both companies will be able to learn from.”

In other words: Facebook may not copy Instagram’s new AI-powered comment filters 1:1, but the company surely is looking to this as a test case as it evaluates if and how it may one day use artificial intelligence for moderation on Facebook proper as well.