Anthropic has switched Claude Security code scanning to run on its most powerful model, Mythos 5, now in public beta for Claude Enterprise customers. An admin enables it in the console; users pick a repository and Claude reports each finding with a CWE category, confidence score, severity rating, and suggested fix.

  • No direct model access: Mythos 5 runs in the background and only findings are exposed — users can’t ask the model for exploits through this interface.
  • Human approval required: patching happens in Claude Code with models the organization already has access to, and every patch must be reviewed and approved by a person.
  • No extra cost: scanning counts as regular token usage under the existing plan; the same embedded-model approach will apply to partner products.

Separately, Anthropic launched the 0xDAF Defender Advantage fund — $35 million in Claude credits for people who fix vulnerabilities in open-source code.

Related: Anthropic Launches Claude Code Security, Anthropic’s Mythos Finds Mathematical Weaknesses in Cryptographic Algorithms

Bringing Claude Mythos 5 to more defenders