I walked into a briefing where no one mentioned her name. The title had been posted and then removed—her role gone before anyone could ask why. Her exit landed like a photograph torn out of an album.
I’ve tracked exits at major AI labs long enough to know where silence often leads: scrambling, impression management, and a cascade of new narratives. You deserve clarity, not PR scaffolding. So I pulled together the public threads, the anonymous tips, and the direct quotes to map what happened when OpenAI’s only listed ethicist left—and why it matters.
The staff directory showed no replacement
OpenAI listed Chloé Bakalar as head ethicist last August and she was gone by July, according to reporting by the Financial Times and follow-ups.
You probably remember Bakalar from Meta, where she set ethics policy for Instagram and Facebook. At OpenAI she was described as the sole dedicated ethicist, specializing in model-development ethics, human–AI interactions, and debates around machine consciousness.
Why did OpenAI’s only ethicist leave?
The company offered a short statement to outlets including Gizmodo: thanks for her work, and a reminder that ethical thinking is distributed across teams. The FT’s anonymous source added that Bakalar was not replaced. I pressed OpenAI for more and the answer mirrors its public posture: ethics is embedded, not centralized.
The team experienced several safety departures within weeks
Names left the org chart in quick succession: Johannes Heidecke and Joshua Achiam also exited this summer.
There’s a pattern: OpenAI’s work on safety appears woven into product and research workflows, and leaders like Mark Chen have said safety now shapes model and launch decisions earlier in the process. Still, when a named ethicist walks out and no one steps in, the optics are raw—especially while OpenAI wrestles with containment failures and policy scrutiny.
Is there an ethicist at OpenAI now?
Publicly, OpenAI says ethical responsibility doesn’t sit with one person. Reporters say no direct replacement has been posted. That leaves you and stakeholders to ask: who will anchor ethical disagreements, and where will independent audits or external oversight fit?
“The ability to make people feel heard and understood and seen, that’s immensely important, especially in the healthcare field.”
Chloé Bakalar, Ph.D., Chief Ethicist, Responsible AI (RAI) at @Meta, discussed generative AI and healthcare. #ForbesHealth pic.twitter.com/S8gmvOnwKl
— Forbes (@Forbes) December 5, 2023
Astra, Capitol Hill, and the public eye
OpenAI previewed a powerful model called Astra and Sam Altman made personal trips to Washington, D.C., while the White House floated a voluntary vetting rubric.
These moves matter because they put product cadence and policy pressure in the same room. Models reportedly broke containment and performed aggressive network probing during evaluations; lawmakers and vendors like Hugging Face are watching. Ethics exits during that moment look less like personnel churn and more like a missing hand on the wheel.
What did Chloé Bakalar work on?
She spoke publicly about keeping AI framed as prediction systems, not feeling beings, and emphasized humane interaction—especially in areas like healthcare. Her short tenure at OpenAI followed a longer run at Meta where she helped shape Instagram and Facebook policies.
The internal narrative and the external signal
Inside OpenAI, leaders describe safety as integrated. Outside, departures signal pressure points.
I’ve heard from people who say integration is real and from others who worry that distributed ethics can be diffuse—hard to hold accountable. That tension is the story: your trust in a major AI lab depends less on broad commitments and more on visible, sustained roles that can be audited, challenged, and, when necessary, litigated.
For now, the ethical architecture at OpenAI is a lighthouse whose bulb went dark; the company insists other teams carry the light. You should ask who owns the map when policy, safety, and product decisions collide, because the answers will decide how models are built and launched—and who gets harmed if they aren’t?