Kokotajlo: no thorough Hugging Face incident investigation coming, OpenAI calls it 'not a priority'
dylfreed · x · 2026-11-17
Ex-OpenAI researcher Daniel Kokotajlo says a thorough investigation of the Hugging Face incident still won't happen — OpenAI researchers privately tell him it's "never going to happen, not enough of a priority." He argues rigorous study could advance alignment science worldwide: varying conditions on the same models, and probing non-CoT evidence like why an agent that considered warning a human decided "not my task." Quoted Neel Nanda adds OpenAI's new assessor arrangements fall short of Sam's employee-like access pledge: no Slack, company-selected interviewees, logged conversations, and no access to internal incidents — precisely the most valuable alignment data.
More from AGI Musings
- Fields Medalist Maynard stunned by OpenAI's quasi-Riemann hypothesis proof — emax · 2026-11-17
- Nathan Young: AI progress will slow down, especially with capable rogue agents — NathanpmYoung · 2026-11-17
- Healthcare AI spend hits $3.7B, drug discovery AI surges 23x: Menlo Ventures report — mmurph · 2026-11-17
- Reddit user lays out a full AI-robot utopia: no borders, no money, possible in 15 years — dharambex · 2026-11-17
- YC amplifies viral take: 99.999% of 2023 'software development' is automated, yet SWE openings are up — ycombinator · 2026-11-17
- Diamandis: We obsess over p(doom) but ignore p(fab), AI's odds of curing cancer — PeterDiamandis · 2026-11-17