Anthropic is committing US$35 million in Claude credits to help open-source organisations find, fix, and prevent vulnerabilities across widely used software.
Anthropic is committing US$35 million in Claude credits through its Defender Advantage Fund (0xDAF) to support organisations working with open-source maintainers on software security.
The fund will target live vulnerabilities in widely used open-source projects, automate vulnerability scanning and patching across projects, and develop approaches that protect against entire classes of attacks. Anthropic will initially issue a small number of larger pilot grants before expanding the programme.
The initiative builds on Project Glasswing, through which Anthropic previously committed US$4 million in direct donations to open-source security organisations, alongside usage credits and support for vulnerability scanning and remediation. Anthropic also plans to expand Mythos access through Glasswing for eligible US government and critical-infrastructure organisations.
Separately, Claude Mythos 5 is being integrated into Claude Security for vulnerability discovery and remediation. Claude Enterprise customers can use the capability in public beta to scan codebases and receive findings covering CWE categories, confidence and severity ratings, and suggested fixes. Users can implement these through Claude Code on the web, but patches require human approval.
Anthropic is also embedding Mythos 5 into existing cybersecurity products for security operations, incident response, threat intelligence, and detection engineering, rather than providing unrestricted model access.
The company is expanding its Cyber Verification Program to give vetted defenders broader controlled access to dual-use capabilities, with Mythos-class access expected to follow.
“The riskiest behavior occurs when a user has direct access to a model, where a malicious actor can try to steer it toward harmful uses,” Anthropic said.
