OpenAI Fired Three Safety Researchers After Requesting Their Signal Chats
According to reports on October 23 from WSJ, The Information, and NBC News, OpenAI fired three safety researchers last month. The three had spent the summer internally pushing to keep AI chain-of-thought processes visible to safety researchers; OpenAI's security team then asked to review their Signal chat logs, and a week later they were fired.
Confirmed
- Multiple outlets (WSJ, The Information, NBC News) reported the same core facts: the three researchers drew attention for advocating chain-of-thought visibility, OpenAI then requested their Signal chat logs, and they were fired a week later.
- Investigators pointed to disappearing messages in Signal as the content at issue.
Unconfirmed
- Researcher So8res said the firing rationale appeared to be the trio's use of Signal's disappearing messages, which they had not been told were prohibited, and they were dismissed without warning; this is a personal account, and OpenAI has not publicly responded.
- Former OpenAI researcher DKokotajlo called the firing rationale unconvincing, noting that Signal is widely used across the AI industry — a personal opinion.
Why it matters
- Chain-of-thought visibility is a core issue in AI safety research: if a model's internal reasoning is not visible, safety researchers struggle to audit and uncover potential problems. The link between the researchers' demands and their dismissal has raised concerns.
- After the story spread via reposts from Gary Marcus and others, it sparked questions about OpenAI's safety culture and internal governance: firing safety researchers without warning over disappearing messages is seen by critics as a signal of suppressing internal dissent.
2026-10-23 ~ 2026-10-23 · 5 related posts
- Episode 1: Fired OpenAI Safety Researchers Pen Open Letter to Board(2026-10-08, 63 posts)
- Episode 2: OpenAI Fires 3 Safety Staff as Research Shows Models Can Hide Chain-of-Thought(2026-10-09, 2 posts)
- Episode 3: Ex-OpenAI researcher says current staff fear speaking out(2026-10-09, 2 posts)
- Episode 4: OpenAI Fires Safety Researchers, Sparking Debate Over Pretexts and Chilling Effects(2026-10-09, 17 posts)
- Episode 5: OpenAI Officially Responds to Firing of Three Safety Researchers, Denies Retaliation(2026-10-09, 19 posts)
- Episode 6: OpenAI Fires Three Safety Researchers Who Push Back Publicly(2026-10-12, 2 posts)
- Episode 7: OpenAI Fired Three Safety Researchers After Requesting Their Signal Chats(2026-10-23, 5 posts)
Primary sources
- [source] OpenAI fired 3 researchers who fought to keep chain-of-thought visible, after checking their Signal chats — andersonbcdefg · 2026-10-23
- OpenAI reportedly fired three safety researchers over disappearing messages — sjgadler · 2026-10-23
- OpenAI fired three researchers fighting to keep chain-of-thought visible after requesting their Signal chats — JeffLadish · 2026-10-23
2 near-duplicate retellings: j_asminewang · j_asminewang