AI & Enterprise
Security experts say zero-day detection already possible with existing AI models after Mythos shock
Security experts say the capability shown by Anthropic’s AI model Mythos in finding thousands of software vulnerabilities can already be replicated using existing models. CNBC reported that watchTowr Labs CEO Ben Harris said public models can be tuned to produce similar results. Vidoc researchers said they detected Mythos-like flaws using orchestration techniques with OpenAI and Anthropic models. Critics also say Mythos access is limited, slowing verification and defenses.
