OpenAI Eliminated the One Person Whose Job Was to Say No

TITLE: OpenAI Eliminated the One Person Whose Job Was to Say No BODY:

The departure was not announced. No press release. No blog post. No LinkedIn update.

Chloe Bakalar's exit took place last month and was not publicly announced, and she has not updated her LinkedIn profile to reflect the departure. That silence is itself a data point. Companies announce departures when they want credit for a transition. They say nothing when there is no story they want told.

Bakalar was OpenAI's only dedicated ethicist, and the company currently has no replacement for the role, though OpenAI maintains that ethics work is embedded across multiple teams involved in developing its models.

Translation: ethics is now everyone's job, which is the corporate formulation for ethics being no one's job.

The Sequence Is Not Random

The consensus reading treats this as routine scaling. The sequence tells a different story.

Between January 2023 and December 2024, OpenAI lost its entire leadership structure dedicated to long-term AI safety. Chief scientist Ilya Sutskever and Jan Leike, who co-headed the "superalignment" team, both resigned. Leike said safety had increasingly "taken a backseat to shiny products" at the company.

That was before 2025. The pace has not slowed.

Caitlin Kalinowski, head of robotics, resigned over the company's agreement with the U.S. Department of Defense. She cited concerns that "surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got."

In July 2026, OpenAI's head of safety systems, Johannes Heidecke, left following an internal reorganization that merges safety and research teams under a single leader. Heidecke was the sixth senior safety leader to depart in two years.

Chief Research Officer Mark Chen wrote: "The demands on safety continue to increase — we are training models at a much faster cadence, and release cycles have come down greatly in turn."

That sentence is the trap. Faster cadence and shorter release cycles argue for more independent safety capacity, not less. The conclusion drawn is the opposite of the stated premise.

Bakalar's exit follows Johannes Heidecke and Joshua Achiam, who also departed in 2026. The pattern points to an organization that has concluded ethics work, as a standalone institutional function with genuine authority to slow or block decisions, costs more than it's worth.

The IPO Problem

This is where the structural risk becomes material.

OpenAI filed a confidential S-1 with the SEC on June 8, 2026, targeting a public debut as early as Q4 2026 or 2027. The company was last valued at $852 billion in a $122 billion funding round closed in March 2026.

Institutional investors doing diligence at that valuation will ask who is responsible for catching problems before they become liabilities. The answer, right now, is an empty box where the ethics function used to be.

OpenAI's public safety narrative has not changed. The org chart has. The most likely explanation is that the company has calculated it can maintain the public narrative through an IPO window without the operational infrastructure to back it. That bet works until an incident forces the comparison into the open. At that point, the gap is not a PR problem. It is a securities disclosure problem.

Anthropic, founded by former OpenAI employees who left over safety concerns, has made responsible AI development a core part of their brand identity. That is the competitive signal OpenAI is sending to every safety researcher on the market right now.

What to Watch

First, whether the public S-1 discloses the current state of ethics governance honestly, or describes a function that no longer exists. The public S-1 amendment filed before the roadshow will be the first look at audited financials and full risk factors. Read the governance section.

Second, whether any departing safety researcher names the lack of dedicated ethics leadership in their exit. None has done so publicly yet. If one does before the roadshow, the timing will be hard for underwriters to ignore.

Third, whether any large enterprise customer or government counterparty asks who at OpenAI is responsible for ethics review of their deployment. The answer is now: the teams building the product.