Meta Disbands Its Responsible AI Team, Moves People to Generative AI

Meta has quietly broken up its Responsible AI (RAI) team. According to a report from The Information, most of the team’s members are being moved to the company’s generative AI product group, while a smaller portion will work on AI infrastructure. The news broke on November 18, 2023, and Meta confirmed the shift to The Verge shortly after.

This isn’t the first time the RAI team has taken a hit. Earlier this year, Business Insider reported that layoffs had already left the team as “a shell of a team.” That report also noted that the RAI team, which was formed in 2019, had very little autonomy. Its initiatives had to go through lengthy stakeholder negotiations before anything could actually be implemented. So in practice, the team was already struggling to exert real influence before this latest restructuring.

Meta’s official line, as told to The Verge by communications rep Nisha Deo, is that this change is meant to “continue to prioritize and invest in safe and responsible AI development.” She added that the RAI members now in the generative AI organization will “continue to support relevant cross-Meta efforts on responsible AI development and use.” I’m not buying it. Moving the oversight team into the product team is like asking the fox to guard the henhouse. You don’t dissolve a dedicated safety unit and then claim you’re still prioritizing safety.

The RAI team was originally created to identify problems with Meta’s AI training approaches—things like whether models are trained with adequately diverse information, with an eye toward preventing moderation issues on its platforms. And there have been plenty of those issues. A Facebook translation error led to a false arrest. WhatsApp’s AI sticker generator produced biased images with certain prompts. Instagram’s algorithms helped people find child sexual abuse materials. These aren’t hypothetical risks; they’re real, documented failures.

This move comes at a time when governments around the world are scrambling to create regulatory guardrails for AI development. The US government has entered into agreements with AI companies, and President Biden directed agencies to come up with AI safety rules. The European Union has published its AI principles and is still struggling to pass its AI Act. Meta’s decision to disband its dedicated safety team sends exactly the wrong signal.

And Meta isn’t alone in this. Microsoft made a similar move earlier this year, restructuring its own responsible AI team. It seems like the pattern is clear: when the pressure is on to ship generative AI features, safety teams are the first to get reassigned or cut. The corporate messaging about “continuing to prioritize responsible development” starts to sound hollow when you see the actual organizational decisions.

I’ve been watching this space for a while, and I can’t help but feel this is a step backward. The RAI team wasn’t perfect, but it existed as a formal check on the company’s AI ambitions. Now that check is gone, absorbed into the very teams that are incentivized to move fast and break things. We’ll see how this plays out, but I’m not optimistic.