On Monday, the Financial Times reported that Chloé Bakalar, OpenAI’s sole dedicated AI ethics lead, had left the company in July after less than a year in the role. She was the third senior safety-adjacent figure to depart in 2026, following safety systems head Johannes Heidecke and former chief futurist Joshua Achiam. The Hacker News thread lit up within hours. The familiar narrative snapped into place: another canary flees the coal mine, another sign that the company is abandoning principle for product.

That narrative is comfortable. It is also wrong — not because the departures don’t matter, but because it mistakes a structural inevitability for a moral crisis. The real story isn’t that OpenAI’s ethicist left. It’s that the role was designed to be unworkable, and everyone involved knew it.

One Person, No Team, No Leverage

Bakalar was, by all available accounts, the only person at OpenAI whose full-time job was AI ethics. Not the head of a department. Not the leader of a team. The sole dedicated ethicist inside a company that employs several thousand people and is valued north of $150 billion.

Think about that ratio for a moment. One person, embedded in an organization where the gravitational pull of product deadlines, revenue targets, and competitive pressure is measured in billions of dollars. What was she supposed to do? Walk into a product review and tell a room of engineers and executives that the thing they’d spent six months building had ethical problems — and expect anything to change? Without a team to run analyses, without a budget to commission external audits, without a reporting line that gave her real veto power?

This is not a criticism of Bakalar, who by all indications is a serious scholar — she holds a Ph.D. from the University of Pennsylvania and has published extensively on the ethics of machine learning. It is a criticism of the structure she was placed in. A single ethicist in a company of thousands is not a function. It is a fig leaf.

The Performative Hire Is a Tech Industry Tradition

The pattern is not new. For years, major platforms have cycled through heads of trust and safety, chief ethics officers, and directors of responsible innovation — often with short tenures and ambiguous mandates. The role is announced with a press release. The hire is celebrated as evidence that the company takes the issue seriously. And then, quietly, the person leaves, and the cycle repeats.

What’s notable about OpenAI’s version of this story is the speed. Bakalar lasted less than twelve months. Heidecke’s safety systems team was restructured into closer proximity with research — which, depending on your priors, either means safety got a seat at the table or safety got absorbed into the very function it was supposed to scrutinize. The company’s framing emphasized the former. The departures suggest the latter.

As one former policy staffer at a rival AI lab put it in a Signal message Monday night: “Nobody takes a job like that expecting to win. You take it hoping you can lose slowly enough to make a difference. The problem is when the company decides it doesn’t even want to lose slowly.”

What Accountability Actually Looks Like

If the lone-ethicist model is broken — and the evidence suggests it is — what replaces it? The comfortable answer, especially in policy circles, is more internal governance: bigger teams, stronger charters, board-level oversight. And those things might help at the margins. But they share the same fundamental weakness: they rely on the company policing itself.

Real accountability for AI systems is unlikely to come from inside the building. It will come from outside — from regulators who can impose fines, from plaintiffs who can bring lawsuits, from insurers who can price risk, and from customers who can walk away. None of these mechanisms requires a head of ethics. All of them require transparency that companies currently resist.

This is the uncomfortable pivot the ethics-departure discourse keeps avoiding. It is easier to mourn the loss of a principled individual than to confront the possibility that the entire internal-ethics model is a category error — a way of looking serious without being serious, of hiring one person to absorb a problem that belongs to the whole organization and, increasingly, to the public.

The Silence That Matters

Bakalar has not publicly explained her departure. Neither has OpenAI, beyond the standard expression of gratitude and well-wishes. That silence is itself a data point. When a company’s sole ethicist leaves after less than a year and nobody will say why, the most reasonable inference is not that everything is fine.

But the question worth asking isn’t “What did Bakalar see that made her leave?” It’s “What did OpenAI expect her to accomplish, and why was the role structured in a way that made accomplishment nearly impossible?” The answer to that question says more about the state of AI governance than any single resignation ever could.

The ethicist is gone. The fig leaf has fallen. The question now is whether anyone — inside the company or out — is prepared to plant something real.

Sources