Anthropic Mythos AI Finds Thousands of Cyber Flaws

Anthropic’s latest AI model, Mythos, has identified thousands of critical vulnerabilities in major operating systems and web browsers ^[1]. Although the company designed the frontier model for defensive security, its discovery of widespread flaws is drawing scrutiny from government officials ^[1].

The model's ability to find exploits at scale has sparked fears about autonomous cyberattacks ^[1]. Analysts suggest that Mythos could enable hackers to launch sophisticated operations with unprecedented speed ^[2].

Meanwhile, Anthropic is investigating reports of unauthorized access to the model ^[3]. This investigation follows concerns that the AI's capabilities might have been misused for hacking ^[3]. Safety advocates are calling for stricter controls on high-capacity AI systems to prevent large-scale digital threats ^[2].