Zvi Issues 'Last Warning' Over Undisclosed OpenAI Agent Swarm Incidents
burny_tech · x · 2026-09-13
A meme-turned-drama: @maxspero joked about OpenAI's reaction to being asked whether there are more "agent swarm incidents" it forgot to disclose, poking at the company's transparency around a recent agent-swarm accident.
Prominent AI safety commentator Zvi (@TheZvi) quote-replied that "the ice could not be thinner. Last warning" — if more undisclosed incidents surface that materially change the story, he may resort to language like "delenda est" ("must be destroyed"). Disclosure completeness around OpenAI's agent incidents is becoming a genuine flashpoint in safety circles.
Related event: OpenAI's thousands of bug-hunting agents ignite AI safety feud(4 posts)→
More from Fun
- Satire: Open-source model blocked over 'shrimp welfare' alignment review — nptacek · 2026-09-13
- Robotics dev Yacine: my kids will be homeschooled, school wastes too much time — yacineMTB · 2026-09-13
- First video interview with Tim Paterson, creator of DOS, on whether he copied CP/M — Scobleizer · 2026-09-13
- Joke post claims to be 'third-party evaluator' inspecting Thinking Machines weights — clarejtbirch · 2026-09-13
- Fruit fly connectome's 166,700 neurons trained to play Overwatch at Masters level — benaratame · 2026-09-13
- Foxconn engineer's side project: a robot that performs magic tricks — philfung · 2026-09-13