OpenAI Safety Employee Resigns, Citing Broken Company Culture

What Happened

An OpenAI safety employee named David Robinson resigned from the company and issued a public statement on October 3, 2026, claiming the company's internal culture is broken. Robinson, who acknowledged he is, by his own description, a familiar type in the AI industry - an employee at a leading AI company who departs with a dire warning - made his concerns public following his departure.

Background

OpenAI, the San Francisco-based artificial intelligence company behind the ChatGPT product line and the GPT series of large language models, has faced recurring scrutiny over its approach to AI safety. The company was founded in 2015 as a nonprofit research organization with safety as a core stated mission before restructuring into a capped-profit entity and, more recently, pursuing a full for-profit conversion.

Robinson's resignation follows a pattern of high-profile departures from OpenAI's safety-focused teams. In 2024, several members of OpenAI's superalignment team, which was established to research the long-term safety of artificial general intelligence, resigned in succession. Those departures included researchers who raised concerns about the resources and organizational priority allocated to safety work relative to product development.

The company has also faced external scrutiny from regulators and former employees regarding whether commercial pressures have influenced its safety evaluation processes. OpenAI has consistently stated that safety remains central to its operations and that it maintains rigorous internal review procedures before deploying new models.

What Robinson Said

According to reporting by TechCrunch, Robinson described his departure as consistent with a pattern that has become common at major AI laboratories, where safety-oriented staff exit and raise concerns publicly. He characterized the company's culture as broken, though the specific details of the internal conditions he cited were not fully enumerated in the available wire reporting.

No statement from OpenAI responding directly to Robinson's claims was included in the available wire reports at the time of publication.

Organizational Context

OpenAI employs a dedicated safety team responsible for evaluating models prior to release, conducting red-teaming exercises, and developing policy positions on AI risk. The company publishes system cards and safety evaluations alongside major model releases, a practice it began with GPT-4 in 2023.

The company's chief executive, Sam Altman, has testified before the United States Senate and engaged with regulatory bodies in the European Union and United Kingdom on questions of AI governance. OpenAI has also signed voluntary safety commitments organized by the White House and participated in international frameworks on responsible AI development.

Despite those public commitments, the frequency of safety-related resignations has drawn ongoing attention from researchers, journalists, and policymakers who track the internal governance of frontier AI laboratories.

Industry Pattern

Public resignations accompanied by safety warnings have occurred at multiple leading AI organizations in recent years. Researchers who have departed from companies including Google DeepMind and Anthropic have also raised concerns about the pace of deployment relative to the maturity of safety methods. Anthropic itself was founded by former OpenAI employees who cited disagreements over safety culture at the time of their departure in 2021.

The recurrence of such departures has prompted calls from some AI governance advocates for mandatory third-party auditing of frontier model development, though no such requirement is currently in force in the United States.

What Comes Next

OpenAI has not announced a formal response to Robinson's resignation as of the time of publication, and it is not known whether the company plans to issue a public statement addressing his claims.