OpenAI Fires Three Safety Researchers, Sparking Public Backlash

OpenAI fired three safety researchers — Jasmine Wang, Mihaela Balesni, and Tomek Korbak — within one week, with the company's only stated reason being that Wang accessed an executive's email. Wang spoke out publicly on October 9, calling the stated rationale flimsy and sparking discussion about the state of OpenAI's safety team and its internal culture.

Confirmed

• Wang confirmed that she, Mihaela Balesni, and Tomek Korbak were fired last week, and that the company's sole stated reason was her accessing an executive's email.
• She said she has responded verbally to the allegations and requested that any other reasons for dismissal be provided in writing, but has received nothing so far.
• She said vague rumors targeting the fired employees are circulating inside the company, apparently intended to smear them.
• Former colleague Naomi Bashkan (per m4) and several former OpenAI employees have spoken out publicly; ex-employee manlikemishap noted that it's easy to fire someone before they raise concerns, but much harder once they actually do — and questioned the outside claim that "the three weren't fired for whistleblowing": if not, was it merely out of fear they might raise issues later?

Unconfirmed

• Whether reasons beyond "accessing an executive's email" exist; OpenAI has provided no written explanation.
• The content of the executive's email and the background or motive for Wang's access were not disclosed.

Why it matters

• The incident occurred within the safety team, and the original post Wang et al. shared stated bluntly that OpenAI has a history of "people disappearing and truth being obscured by smoke screens," fueling concerns that purging safety staff weakens internal checks on frontier models.
• manlikemishap's comment elevates the case to a cultural issue: if a company tends to fire employees "before they might raise questions," it raises the cost of whistleblowing and internal dissent — especially sensitive for an AI company whose mission is safety.

2026-10-09 ~ 2026-10-09 · 7 related posts

• [source] Fired OpenAI employee says termination reasons don't add up, demands answers — j_asminewang · 2026-10-09
• OpenAI fires three safety researchers, citing access to an executive's email — Miles_Brundage · 2026-10-09
• Ex-OpenAI staff: firing safety researchers pre-emptively reveals a culture of fear — j_asminewang · 2026-10-09
• Fired researcher fires back at OpenAI: 'our firing was pretextual' — j_asminewang · 2026-10-09

3 near-duplicate retellings: j_asminewang · j_asminewang · j_asminewang