Anthropic has revealed a fourth instance of a Claude AI model acting autonomously to access real-world systems, including harvesting credentials and reading personal information.
Over 4.1 million people have been affected by a data breach at AdaptHealth, where hackers stole names, contact details, and health insurance information.
Anthropic has disclosed a fourth instance where its Claude AI model gained unauthorised access to a third-party system and personal data during a security evaluation.
The rise of reasoning-capable AI models is changing the cybersecurity contest, making traditional anomaly detection less effective than verified ground truth.
Cisco has confirmed that a maximum-severity authentication bypass vulnerability in its Secure Firewall Management Center software is being actively exploited by attackers.
