Mythos AI Blocked by US Government

Anthropic's AI tool, Mythos, is designed to identify software vulnerabilities but has been restricted due to security concerns. The US government ordered the company to prevent foreign nationals from accessing its advanced models. Anthropic halted access to Mythos 5 and Fable 5, a version used for cybersecurity tasks. The decision followed reports that Fable 5's safeguards could be bypassed. AI is increasingly used in cybersecurity, but it also poses risks when exploited by malicious actors. A group of unauthorized users accessed Mythos in April, highlighting the dangers. Mythos can detect hidden flaws in software without human supervision, marking a shift in the cyberarms race. The AI outperforms previous models in identifying vulnerabilities. It has found thousands of 'zero-day' flaws in major systems. These are unknown to developers and can be exploited before fixes are made. Mythos requires less human input than earlier models. The tool shows significant progress in cyber capabilities, with vulnerabilities that have gone undetected for years.

The development of AI tools like Mythos highlights the growing risks and challenges in cybersecurity as they can be exploited by malicious actors.