Anthropic is broadening access to the cybersecurity capabilities of its advanced AI models through a mix of partner integrations, an updated Claude Security offering, a new open source funding program, and plans to expand its Cyber Verification Program. The move builds on Project Glasswing, launched in April, which gave a small group of organizations early access to Claude Mythos Preview and its successor, Mythos 5.<\/p>\n
Anthropic said the goal of its earlier access program was to give defenders time to find and fix vulnerabilities before comparable capabilities became widely available or fell into the hands of malicious actors. Claude Fable 5 followed as a broadly available model that keeps dual-use cyber work blocked. Anthropic said the riskiest scenario is direct, unrestricted access to a model, a risk that drops sharply when users instead receive specific defensive outputs, such as a patch or a security alert.<\/p>\n
The latest changes are built around this idea, expanding what defenders can get from Mythos-class models while keeping guardrails around direct interaction with them.<\/p>\n
On the integration front, Anthropic is working with cybersecurity partners to build Mythos 5 into the security operations, incident response, and detection tools already used by teams protecting hospitals, utilities, financial systems, and the software supply chain.<\/p>\n
End users will not interact with Mythos directly. Instead they will work through purpose-built interfaces that run the model in the background and return only a defined output, such as a list of suggested patches, with abuse-prevention checks meant to keep the model within that scope.<\/p>\n
Claude Security, currently in public beta for Claude Enterprise customers, now runs its codebase scans on Mythos 5. Scans surface each finding with a CWE category, confidence and severity ratings, and a suggested fix. Any fix still has to be implemented through Claude Code and approved by a human before deployment.<\/p>\n
Anthropic is also launching a $35 million open source security fund to back projects that defend the open source ecosystem. The fund is described as the largest of its kind and will support a mix of established and emerging projects focused on identifying and remediating vulnerabilities in open source software.<\/p>\n
Open-weight models, in Anthropic’s view, are an important part of the AI ecosystem, and the company wants to support efforts to keep the open source software stack secure. Specific recipients and grant amounts were not detailed in the announcement.<\/p>\n
Anthropic’s Cyber Verification Program, which audits the cyber capabilities of its frontier models, is being expanded. The expansion is aimed at giving security teams and policymakers more visibility into how Anthropic’s models perform on offensive and defensive cyber tasks.<\/p>\n
The combined push, partner integrations, an upgraded Claude Security beta, a dedicated open source fund, and broader verification work, is designed to widen the defensive surface available to security teams while keeping direct access to Mythos-class models restricted.<\/p>\n
Mythos 5 is the successor to Claude Mythos Preview, part of Anthropic’s Mythos-class AI line introduced through Project Glasswing in April. It is offered to defenders through partner integrations and the Claude Security beta rather than through direct, unrestricted user access.<\/p>\n
Anthropic announced a $35 million open source security fund to support projects that identify and remediate vulnerabilities in open source software. The company described it as the largest fund of its kind.<\/p>\n
Claude Security, in public beta for Claude Enterprise customers, now runs its codebase scans on Mythos 5. Each finding includes a CWE category, confidence and severity ratings, and a suggested fix, with implementation routed through Claude Code and requiring human approval before deployment.<\/p>\n