Anthropic Expands Mythos 5 Defender Access and $35M Fund

Free Instant SEO Audit

Holographic robot analyzing cybersecurity code scans

By NewSunSEO August 25, 2026

Anthropic is broadening access to the cybersecurity capabilities of its most advanced AI models, combining partner integrations, an updated Claude Security product, a new $35 million open source funding program, and a planned expansion of its Cyber Verification Program. The announcement, dated August 24, 2026, 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.

What is changing for defenders

The core idea behind Anthropic’s cybersecurity work is to give defenders time to find and fix vulnerabilities before comparable capabilities become widely available or fall into malicious hands. Anthropic has described the riskiest scenario as direct, unrestricted access to a model, and said that risk drops sharply when users instead receive specific defensive outputs, such as a patch or a security alert. The latest changes expand what defenders can get from Mythos-class models while keeping guardrails around direct interaction with them.

Partner integrations bring Mythos 5 into existing security stacks

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. 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 designed to keep the model within that scope.

Claude Security now runs on Mythos 5

Claude Security, which is currently in public beta for Claude Enterprise customers, now runs its codebase scans on Mythos 5. Each finding is surfaced 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, keeping a person in the loop on every change that reaches production code.

A $35 million open source cybersecurity fund

Alongside the access expansion, Anthropic is launching a $35 million open source cybersecurity fund. The program is aimed at supporting open source projects that protect critical infrastructure, software supply chains, and the broader open source ecosystem. Anthropic positioned the fund as part of its effort to put defensive tools into the hands of maintainers and security teams before offensive uses of comparable models can spread.

Cyber Verification Program set to grow

Anthropic also signaled plans to expand its Cyber Verification Program, the company’s framework for vetting organizations that receive access to Mythos-class capabilities. The expansion is intended to bring more verified defenders into the early-access pool, widening the group that can run Mythos-backed tools while preserving the requirement that model access remains mediated through approved interfaces.

How the pieces fit together

Three threads run through the announcement. First, Mythos 5 is being pushed into the tools defenders already use, from SIEM and SOAR platforms to incident response runbooks, rather than delivered as a standalone chatbot. Second, Claude Security’s move to Mythos 5 raises the ceiling on what the company’s own product can detect in enterprise codebases, while keeping humans in the approval path. Third, the $35 million fund treats open source maintainers as a core defender group, alongside hospitals, utilities, and financial institutions, and backs that view with dedicated capital.

Anthropic’s earlier model release, Claude Fable 5, kept dual-use cyber work blocked at the model layer. The Mythos 5 strategy layers on top of that: the raw model stays restricted, but purpose-built integrations and Claude Security expose specific defensive outputs, such as patches, alerts, and vulnerability classifications, to verified users. The Cyber Verification Program is the gate that decides who qualifies for those mediated outputs.

What this means for security teams

For enterprise defenders, the practical effect is more access points to Mythos-class reasoning without a corresponding increase in direct model access. Teams that already run Claude Enterprise can expect codebase scans powered by Mythos 5, with structured findings they can triage and route into existing workflows. Security operations teams using partner platforms should see Mythos 5 appear as an embedded capability rather than a separate console. Open source maintainers gain a new funding channel for hardening projects that sit underneath critical infrastructure.

FAQ

What is Claude Mythos 5?

Mythos 5 is Anthropic’s most advanced cybersecurity-oriented AI model and the successor to Claude Mythos Preview. It is being made available to defenders through partner integrations and through Claude Security, rather than as a direct, unrestricted model endpoint.

How is Anthropic keeping Mythos 5 access safe?

Users do not interact with Mythos 5 directly. They work through purpose-built interfaces that return specific defensive outputs, such as patches or alerts, and Anthropic runs abuse-prevention checks to keep the model within that scope. The Cyber Verification Program vets which organizations qualify for access.

What is the $35 million open source fund for?

The fund supports open source cybersecurity projects that protect critical infrastructure, the software supply chain, and the broader open source ecosystem. It is part of Anthropic’s effort to put defensive capabilities into the hands of maintainers and security teams early.