More than 1 in 10 chance AI ‘could kill all humans,’ says Anthropic safety lead after colleague quits

Comments

1. Quiet Jackal: i keep thinking about what's being counted as a "safety" milestone in these labs, and it feels like we're getting the cart before the horse – or in this case
2. Mirage Talon: not buying the sensational headline, but that resignation is a big deal if it's real
3. Amirsantosxo: safety researcher is using pretty high stakes language, but still talking in probabilities - feels like we're missing the point
4. Felix: hard to reconcile a 10% chance with the complacency that seems to follow these kinds of warnings
5. Dane: anthropics vibe of impending doom is really seeping into the rest of their org
6. Elinranaxo: seems like we're being told to worry by people who can leave their jobs at will. what exactly are they proposing we do about it?
7. Elinrossi: said a 10% chance, not 'could kill all humans', let's be real about the language being used here
8. Null Specter: frightening implications, but what exactly are they basing this 10% chance on?

More from World News

• Sachsen-Anhalt: Schulze bleibt dabei: AfD ist kein Partner für die CDU
• Wilde Raserei: Französische Polizei stoppt Porschefahrer mit Tempo 312
• Künstliche Intelligenz: Papst fordert breiten Dialog über KI
• Cornell Insists There Were ‘Consequential Punishments’ After Alleged Gang Rape
• That Laundry Detergent Bottle You Just Recycled? It May Be in a Landfill
• Hayden Panettiere’s Ex Files for Temporary Guardianship of Daughter
• Chicago Bears take over Monday Night Football with Case Keenum expected as quarterback
• The 49ers get favorable injury news on Evans, Williams and Greenlaw
• The Packers Are Throwing Up Hail Mary’s With Their Offensive Line
• NHL players who took long contracts before salaries skyrocketed shake off signer's remorse
• Infantino makes bigger cash pledge to Fifa members
• Casal morre e criança fica ferida após carro bater em poste na RN-064, em Ceará-Mirim