Source item

Exclusive: Anthropic wants to pay hackers to find model flaws (leveraging @hackerone) https://t.co/BghExqCxSv