Anthropic has opened limited access to the Claude Mythos Preview model — it autonomously searches for software vulnerabilities and has already found thousands of them.
Anthropic has opened limited access to its new AI model Mythos, designed to locate critical vulnerabilities and counter cyber threats. The tool will only be available to select corporations, as its hacking capabilities are deemed too dangerous for public release. According to The Wall Street Journal, around 50 companies and organizations supporting critical IT infrastructure will receive the preliminary version of the neural network.
