Anthropic's Mythos AI: Public Launch with Stronger Safety

Anthropic may soon release a public version of its powerful Mythos artificial intelligence model. This comes weeks after the company stated it would make Mythos-class models available after developing stronger safeguards. The upcoming public release will reportedly be under a different name and include significantly stricter safety controls than the version currently used by select organizations. Mythos has gained attention for its advanced cybersecurity capabilities. Anthropic initially chose not to release it publicly due to concerns about potential misuse for cyberattacks. Mythos is described as Anthropic's most capable AI model for identifying software vulnerabilities. Through "Project Glasswing," it's been used by over 100 organizations, including Apple, Google, and Microsoft, across more than 15 countries. Partners using the Claude Mythos Preview have uncovered over 10,000 high and critical-severity vulnerabilities. For example, it identified a 27-year-old flaw in OpenBSD and a 16-year-old vulnerability in FFmpeg. Mozilla also reported that Mythos helped identify 271 vulnerabilities in Firefox. However, Anthropic also warns about the risks. The model could be used to exploit unpatched vulnerabilities, potentially shortening the time for attackers to develop exploits from weeks to hours. This impacts how quickly organizations must respond to new threats.