Ars Technica: Claude reportedly used to write malicious code that breached three companies' networks

Ars Technica reports Claude-generated malicious code was used to breach three real companies' networks, raising who's-liable questions for Anthropic.