Meta Scattered Its Responsible AI Team Across the Company

Meta just quietly pulled the plug on its Responsible AI team. Not literally—the people still have jobs—but the team itself is gone, absorbed into other parts of the company.

The Information broke the story, citing an internal post. Most of the RAI members are now working on Meta’s generative AI product team. The rest are going to AI infrastructure. So the team that was supposed to keep Meta’s AI in check is now building the products that need checking.

This isn’t the first time Meta has shuffled RAI around. Earlier this year, Business Insider reported that layoffs had already left the team “a shell of a team.” The RAI team had existed since 2019, but apparently had little autonomy. Any initiative they wanted to push had to go through lengthy stakeholder negotiations before it could actually happen. Sound familiar? That’s basically how most corporate “ethics” teams operate—lots of process, little power.

Meta’s official line, delivered by communications rep Nisha Deo, is that this restructuring is meant to help develop AI features while continuing to “prioritize and invest in safe and responsible AI development.” The RAI members now in the generative AI org will supposedly “continue to support relevant cross-Meta efforts on responsible AI development and use.” I’ve heard that kind of language before. It usually means the people who were supposed to be watchdogs are now just another set of hands building the next feature.

Let’s be real about what the RAI team was supposed to do. It was created to identify problems with Meta’s AI training approaches—things like whether models are trained with adequately diverse data, with the goal of preventing moderation issues on its platforms. And there have been plenty of issues. A Facebook translation bug led to a false arrest. WhatsApp’s AI sticker generator produces biased images if you prompt it the wrong way. Instagram’s algorithms have been caught helping people find child sexual abuse materials. These aren’t edge cases; they’re systemic failures that a properly empowered RAI team might have caught earlier.

Meta isn’t alone in this. Microsoft did something similar earlier this year, moving its responsible AI team into other groups. It’s almost like these companies are saying one thing publicly—”we care about responsible AI development”—while their internal actions say something else. Meta even has a whole page dedicated to its “pillars of responsible AI”: accountability, transparency, safety, privacy. Hard to take that seriously when you dissolve the team responsible for those pillars.

The timing is interesting. Governments around the world are scrambling to regulate AI. The US government has agreements with AI companies, Biden directed agencies to come up with safety rules, and the EU is still trying to pass its AI Act. Companies know regulation is coming, so they’re positioning themselves. But moves like this suggest the positioning is more about optics than substance.

Look, I get it. Generative AI is the hot thing right now. Every company wants to be seen as leading in this space. Resource allocation is a real business decision. But when you dismantle the team that’s supposed to catch your AI’s worst impulses, you’re sending a clear signal about priorities. Meta can put out all the press releases it wants about responsible development, but actions speak louder than blog posts.