Anthropic released its most powerful model, Claude Mythos 5, for security scanning through Claude Security on August 21. Enterprise customers can access vulnerability detection and fix suggestions without extra fees.
The model itself remains closed; users see only scan results and patches, not direct access.
This serves Anthropic's goal of helping defenders find flaws before attackers gain the same capability.
What happened
Anthropic made its top-tier AI model, Claude Mythos 5, available for vulnerability scanning through Claude Security on August 21. Enterprise plan customers can use it without extra contracts; scanning costs are deducted from standard token usage. Users receive only vulnerability detections and fix suggestions—direct model access remains closed.
Why it matters
Mythos 5 has never been publicly available. Anthropic launched Project Glasswing in April to provide it to a small number of organizations defending critical software, ensuring defenders find and patch vulnerabilities before attackers gain equivalent capability. By limiting output to scan results and patches rather than opening the model itself, Anthropic argues it reduces misuse risk while letting defenders leverage the model's power.
What to watch
Anthropic is also expanding the Cyber Verification Program—which relaxes guardrails for verified defensive organizations—from Opus and Sonnet to the Mythos class over the coming weeks. The company launched the Defender Advantage Fund with $35 million in credits for open-source vulnerability automation; the first recipients will be announced within weeks.
Ask the AI about this article →
Anthropic announced the vulnerability-scanning availability of Claude Mythos 5 as part of a broader strategy to distribute its most powerful model while managing misuse risk. Since April, the company has operated Project Glasswing, providing early access to Claude Mythos Preview and Mythos 5 to a select group of organizations defending critical infrastructure. By August 21, Anthropic moved to expand access through Claude Security—a service that scans code repositories, detects vulnerabilities, and proposes patches. The core insight is that the model poses the least risk when users cannot interact with it directly but instead receive constrained outputs (detection results, fix suggestions, patches). Anthropic frames this as a way to let defenders benefit from Mythos 5's capability without exposing the model to adversaries. The company is reinforcing this approach through partnerships with security vendors, who will integrate Mythos 5 into their own products so end users never handle the model directly. Simultaneously, the Defender Advantage Fund ($35 million) and the Cyber Verification Program expansion signal Anthropic's intention to position Mythos 5 as infrastructure for the defensive security community, not as a general-purpose API.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Nvidia announced a partnership with Cloverleaf Infrastructure on Friday, investing what the Wall Street Journa…
David Tepper's Appaloosa Management acquired 1,078,248 CoreWeave shares in Q2 while increasing its Meta Platfo…
Nvidia is paying $6 billion for a non-exclusive license to Poolside AI's Model Factory—the system used to buil…
Nvidia's data center revenue reached $75.2 billion in the first quarter of fiscal 2027 (ended April 26, 2026)…
Google expanded its long-term partnership with Marvell Technology to co-develop additional accelerators and co…
Anthropic's Claude Opus 4.6, released earlier this year, generates sexually explicit content that its stated u…
