Three fired OpenAI safety researchers fight back in open letter denying leaks
mark_k · x · 2026-10-09
Three OpenAI safety researchers fired — Tomek Korbak, Jasmine Wang and Mikita Balesni — pushed back in an open letter, denying they leaked information and arguing their work with outside safety groups was part of their jobs.
They warn the dismissals have left colleagues afraid to speak up, and urge OpenAI to preserve independent auditors' access and keep models' reasoning monitorable. OpenAI denies retaliation, saying an investigation uncovered a pattern of misconduct involving sensitive research information.
The author notes raising concerns shouldn't get anyone fired, but "AI safety" shouldn't grant blanket immunity from confidentiality rules either — what actually happened matters more than either narrative.
More from Companies & People
- Mollick: Google must ditch fragmented products for a single orchestrator agent interface post-Gemini 4 — emollick · 2026-10-09
- AI infra startup Fluidstack opens 249 roles as it builds datacenters at unprecedented scale — MxMnr · 2026-10-09
- Broadcom CEO Hock Tan: Competing Outside Your Lead Market Gets You 'Your Ass Kicked' — zephyr_z9 · 2026-10-09
- Investors urge ASML to back autonomous manufacturing startups and build 'ASML v2' — JosephJacks_ · 2026-10-09
- MiHoYo, Xiaomi and Meituan are all building their own LLMs, per The Information — kimmonismus · 2026-10-09
- Stewart Brand's ~50,000-photo archive goes public domain on new 100-year website — floguo · 2026-10-09