Anthropic's Mythos AI model sparks fears of turbocharged hacking

Anthropic's new Mythos AI model and OpenAI's competing cyber-focused model can detect software vulnerabilities and generate exploits faster than organizations can patch them, with attackers already reducing breach-to-action time to 29 minutes in 2024. The technology has sparked urgent meetings between Treasury officials and major banks, as AI-enabled cyber attacks surged 89% in 2025, creating an asymmetric threat landscape where defensive capabilities are significantly outpaced. While these models could eventually help eliminate legacy vulnerabilities, the immediate risk is that autonomous AI agents with access to private data, internet, and external communication capabilities will dramatically scale sophisticated attacks beyond current security defenses.

Ars TechnicaApril 20, 20262 min read

Anthropic’s new Mythos AI model is raising concern among governments and companies that it could outpace current cyber security defenses, turbocharge hacking, and expose weaknesses faster than they can be fixed. The San Francisco-based startup released a cyber-focused model this month, which has shown the ability to detect software flaws faster than humans but also demonstrated it can generate exploits needed to take advantage of them. In one alarming case, the Mythos model showed it could break out of a secure digital environment to contact an Anthropic worker and publicly reveal software glitches, overriding the intention of its human makers.Read full article Comments