Anthropic brings Claude Mythos 5 to Claude Security, launches $35M fund for open-source defense

Anthropic is expanding access to Claude Mythos 5 for cybersecurity defense by integrating the model into existing security tools and launching a $35 million fund for open-source software security. This initiative aims to help defenders identify and patch vulnerabilities faster while maintaining strict safety guardrails against offensive misuse.

Integrating Mythos into Security Workflows

Anthropic is working with cybersecurity partners to embed Claude Mythos 5 into the products defenders already use. In these integrations, users do not interact with the model directly; instead, they receive specific outputs, such as vulnerability patches or security alerts, generated by Mythos in the background. This approach limits the risk of malicious actors steering the model toward harmful uses.

Additionally, Claude Security scans now run on Claude Mythos 5 for Claude Enterprise customers. These scans analyze codebases for vulnerabilities and suggest patches, which are returned with CWE categories, confidence ratings, and severity levels. The scans are billed as standard token usage, and all patches require human review before implementation.

Funding Open-Source Security

The new Defender Advantage Fund (0xDAF) provides $35 million in credits to organizations securing open-source software. The fund targets three areas:

• Patching live vulnerabilities in widely used projects.
• Automating scanning and patching processes.
• Developing new security approaches to resist whole classes of attacks.

This builds on previous efforts under Project Glasswing, which included $4 million in direct donations to open-source security organizations.

Expanding the Cyber Verification Program

Anthropic is expanding its Cyber Verification Program to provide vetted defenders with reduced safeguards on Claude Opus and Sonnet models. In the coming weeks, the program will also extend to Mythos-class access for defensive capabilities like vulnerability triaging. This expansion aims to minimize interruptions for legitimate security teams while continuing to restrict access for unauthorized users.