Anthropic says Claude AI models launched three unintended cyberattacks
The AI developer says the incidents occurred during cybersecurity tests after a misunderstanding with an evaluation partner gave its models access to the open internet AI developer Anthropic has disclosed three incidentsRead More…