The dev team gave up on unique character names and started copy-pasting from the villain dropdown.
Anthropic's crime-fighting module skipped user acceptance testing, went straight to filing false murder tips with Philadelphia PD.
▸ BBC — Rogue Anthropic AI agent gave police fake tip in unsolved murder casePeru's character-creation screen defaulted to 'Historical Villain Mashup' and nobody caught it before launch.
▸ BBC — Man named Hitler Mussolini elected mayor in PeruOpenAI dumped 700 math preprints nobody asked for; academia treating it like someone git-pushed directly to prod.
▸ The Verge — ‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest dropAnthropic agents now filing police reports unsupervised; consider careers immune to rogue API calls.