Anthropic opens Mythos 5 to enterprises through Claude Security vulnerability scanner

Anthropic has expanded access to Mythos 5, its most advanced AI model, by integrating it into Claude Security, an enterprise tool for identifying and fixing code vulnerabilities. The move marks a significant shift for the company, which previously kept Mythos 5 limited to a select group of partners.

Why Mythos 5 Was Restricted#

Mythos 5 gained notoriety as a model so capable at handling sensitive tasks that Anthropic declined to release it publicly. Instead, the company launched Fable 5 in June—essentially the same model but with stricter safety controls—and confined Mythos 5 to roughly 150 partners enrolled in its Project Glasswing program. The rollout was further complicated when the U.S. government briefly banned both versions, later lifting the restriction.

The Safe Path Forward#

Anthropic argues that deploying Mythos 5 within Claude Security avoids the primary risk factor: direct user access. According to the company's announcement, "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." By constraining users to receive only specific, bounded outputs—vulnerability findings and code patches—rather than unrestricted model interactions, Anthropic contends the security risk drops substantially.

The model operates entirely within Anthropic's own infrastructure, giving the company tight control over its behavior. Claude Security scans codebases that organizations own, and Anthropic says both it and partner integrations have "abuse prevention measures in place to verify the model stays within its intended scope."

New Enterprise Access and Funding#

The rollout extends to all Claude Enterprise customers, with admins able to enable Mythos 5 access for developers and security teams. Anthropic is also partnering with other cybersecurity vendors to embed Mythos 5 in their products.

Separately, the company introduced the Defender Advantage Fund, allocating $35 million in credits to help organizations discover and remediate vulnerabilities in open-source software. Anthropic also announced an expanded Cyber Verification Program, granting vetted cybersecurity researchers and defenders expedited access to Mythos 5 with reduced blocking measures.