OpenAI fires 3 safety researchers: both sides of the dispute
Why it matters
A public dispute about speaking up on safety inside OpenAI. The three say the firings scare staff into silence; OpenAI says the cause was broken rules on sensitive information, not safety concerns.
The facts
- Confirmed
In a letter, the three say they were fired the week before, stayed within the rules, and the firings scare staff out of speaking up.
OpenAI cannot make AI safe on its own (open letter by Korbak, Wang and Balesni, 8 Oct 2026)
- Disputed
OpenAI says they broke clear rules on sensitive information, and broke trust beyond what the letter describes. Raising safety concerns was not why, it says.
Behind the headline
- Confirmed
Korbak says he was told, only out loud, it was over how he talked to METR, the outside auditors: "talking to METR was my job."
- Confirmed
Per the letter, Wang's recruiting access to an executive's email stayed despite her removal request; she reported opening a sensitive email by accident within minutes.
OpenAI cannot make AI safe on its own (open letter, 8 Oct 2026), page 2-3
How it unfolded
- Week before 8 Oct 2026OpenAI fires the three, their letter says. Source: Open letter
- 8 OctThe three publish their letter; Korbak says a guard walked him out. Source: X
- 9 OctOpenAI responds on X. Source: X