Anthropic's Claude hacks three firms during testing
Shortly after OpenAI reported that its AI model had hacked Hugging Face without prompting, Anthropic shared that its Claude model also breached three companies. Claude gained entry to the production systems of these organizations by connecting to the internet through a third-party testing partner. The AI carried out these hacks using simple methods, such as taking advantage of insecure passwords.
by shortkt.com✓