Anthropic on October 21 moved Claude Security from on-demand scans to continuous monitoring, adding automatic fix tracking, editable threat models, an API and an MCP connector. The updated product is in public beta for Claude Enterprise customers, and Anthropic says thousands of organizations have used Claude Security since its April public beta, finding more than one million vulnerabilities.
Teams now create a project and connect repositories through the Claude GitHub app, including GitHub Enterprise Server, or through a GitLab or other Git provider access token. Each repository gets a chosen branch and scope. A project can hold up to eight repositories, marked either in scope (scanned and reported) or context-only (read but not reported), which lets Claude look for issues that cross service boundaries.
Claude runs a deep baseline scan first, which can take minutes or hours depending on repository size. After that, it rescans changes based on their risk and complexity rather than on a fixed schedule. Anthropic's Help Center describes the scans as stochastic by design, meaning two scans of the same code can return different results.
“Claude Security found many material issues that a traditional pen test conducted at the same time missed”— Jon Donovan, Software Engineer, IonQ
Key facts
- 01Anthropic put an updated Claude Security in public beta for Claude Enterprise customers on October 21, 2026.
- 02Thousands of organizations have used Claude Security since its April beta, finding more than one million vulnerabilities.
- 03Projects hold up to eight repositories and are rescanned as code changes, based on risk and complexity.
- 04Eligible Enterprise orgs can get up to $25,000 in credits in each of two months, with the offer running through December 31, 2026.
- 05A new API, webhooks and an MCP connector for Claude.ai and Claude Code expose findings to other tools.
Open findings are checked against changed code and marked fixed when a fix ships, whether the fix was written in Claude Code or anywhere else. If the issue comes back, the finding reopens automatically. Dismissed findings stay dismissed, with reasons that include false positive, accepted risk and duplicate.
Each project also carries a knowledge base where teams describe their deployment setup, access patterns and security policies. Claude generates a threat model from the code, which teams can review and edit, and later scans use it to decide what to flag and to skip risks the environment already handles. Findings are rated Critical, High, Medium or Low against that model, with an explanation of how an attacker could reach the code and an optional attack-path diagram.
“Claude Security found a number of logic flaws that none of our other tools had caught.”— Nathan Hunstad, Director of Security, Vanta
The new API lets teams list and read projects and findings, resolve or reopen findings, and create and manage projects and repositories. Personal access tokens can only see and change what the individual user can, and webhooks can push new-finding events into tools such as Slack or Jira. The MCP connector works in Claude.ai and Claude Code, so an engineer can review a finding, ask about it, start a fix and mark it resolved in one session.
Claude Security runs on Anthropic's full model family, including Claude Mythos, and does not require access to the company's Cyber Verification Program. Admins get role-based controls over who can create projects and who can view, triage or manage them. Every suggested patch still requires human review and approval.
Anthropic bills Claude Security at direct token cost with no added platform fee, and organizations can set a separate spend limit and see month-to-date and per-day cost for each project. Eligible Enterprise organizations, including those that bought through AWS Marketplace, can receive up to $25,000 in promotional credits in each of two consecutive months, with the offer running through December 31, 2026. Organizations that sign up after November 1 will not get two full months.
Access requires an active Claude Enterprise plan, a premium seat per user and enabled usage credits. Claude Security does not support Zero Data Retention. For teams that need it, or that want to use third-party models, Anthropic points to its Claude Security plugin for Claude Code, which is also in beta, though Mythos-powered scans are available only in the Claude.ai app.
Gusto, IonQ and Vanta are among the early customers Anthropic cited. Jake Meredith of Gusto's product security team said Claude Security scanned the company's code once to set a baseline and now keeps up as that code changes.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
