Sarcastic 'AI safety hall of fame' calls out Microsoft's Sydney, OpenAI eval monitoring and xAI
gleech · x · 2026-09-20
A sarcastic "AI safety hall of fame" lists: the Microsoft engineer who forgot system prompt repetition for Sydney (early Bing Chat), the OpenAI person who decided monitoring sandboxed evals wasn't worth it, and the entire 2025 xAI staff — riffing on well-known AI safety lapses.
More from Fun
- "AIs that won't use developers will be replaced by AIs that will," quips X user — ashishllm · 2026-09-20
- Radiohead "predicted" the Hugging Face incident back in 2003 — birchlse · 2026-09-20
- EA community debate: saved lives vs. 'proof by assertion' reasoning style — mjdramstead · 2026-09-20
- Carmack praises Musk for going all-in on SpaceX and Tesla with real bankruptcy risk — elonmusk · 2026-09-20
- KlausCodes Announces Dropping Out of University, Recalls Flying to Canada on Internship Savings — KlausCodes · 2026-09-20
- Five pieces of 25% evidence add up to 76%? Lawyer's probability take sparks Bayesian correction — dilanesper · 2026-09-20