Wholestory

← Voices

Jan Leike

72%credibility0 / 0 / 0

Why this score Former co-lead of OpenAI's Superalignment team (ex-DeepMind); resigned May 2024 citing safety deprioritization. Credible firsthand insider on lab safety posture, with a critical viewpoint of his former employer.

Tracked statements

1
? unverified
over the past years, safety culture and processes have taken a backseat to shiny products

The resigning Superalignment co-lead’s public characterisation of his former employer’s priorities; an insider allegation, recorded as an attributed statement. Corroborated as accurately quoted by Wired and CNBC; the underlying judgement is contested by OpenAI.

state of ai/safety·source ↗