We are committed to accuracy. This is a public record of all corrections made to Ethical AI Departures content.
Changed Steven Adler’s departure date from an unsupported exact day to year precision. In parliamentary testimony, Adler described leaving OpenAI in late 2024. His departure year and evidence classification are unchanged.
Corrected the criteria attached to Geoffrey Hinton’s extinction-risk forecast. The earlier criteria treated existential risk or near misses as qualifying evidence. The forecast concerns extinction itself. His probability estimate and time horizon are unchanged, and the record remains open and excluded from binary accuracy scores.
Corrected the summary of Suchir Balaji’s fair-use essay. The earlier summary included examples involving stock photography and copywriting that the essay did not establish. The revised summary reflects his case-specific analysis of ChatGPT and market substitution.
We completed a sitewide audit of Context only profiles. Three records with person-level evidence were reclassified as Explicitly stated or Reported connection; 26 unsupported person records were archived; and one record was returned to draft pending primary-source verification. Organizational team dissolutions and reorganizations now appear in a separate record. The review changed the headline count from 69 published person records to 39 evidence-linked departures, with unresolved allegations shown separately. Jonathan Richard Schwarz's subsequently added, evidence-linked record brings the current count to 40.