Anthropic has launched Claude Security in public beta, opening a code-scanning product to its enterprise customers and lining up the largest names in cybersecurity and consulting as launch partners. The tool went live yesterday inside the claude.ai sidebar for enterprise tier users, with Team and Max plans next in the rollout queue.
The product scans a codebase the way a human security researcher would, tracing data flows, mapping read sources and tracking how components interact across files before returning an impact-style report on vulnerabilities with reproduction steps and recommended fixes. Customers connect a GitHub repository, let Claude scan it, then drop the resulting report into Claude Code to patch the issues automatically.
Jaeden Schafer said on the podcast the workflow mirrors what Anthropic already does with Claude's design tooling, where one product produces a handoff document that another executes against. "I really love this. I think this is this is a really smart move by anthropic, they don't need an API, they don't need to like have any sort of like, literal integration, it just creates a document and an MD file, right, that just explains what needs to happen." The beta also adds scheduled scans, the ability to dismiss low-priority findings, CVSS scoring and a markdown export that routes straight into ticketing systems.
Key facts
- 01Anthropic launched Claude Security in public beta, available immediately to Claude enterprise customers via the claude.ai sidebar, with Team and Max plans next in line.
- 02Launch partners include CrowdStrike, Palo Alto Networks, SentinelOne, Trend Micro and Wiz, all integrating Claude Opus 4.7 directly into their platforms.
- 03Accenture, BCG, Deloitte, Infosys and PwC are spinning up deployment practices around the tool.
- 04The beta adds scheduled scans, dismissible findings, CVSS scoring and a markdown export that routes into ticketing systems.
The partner list reads like a roll call of enterprise security. CrowdStrike, Palo Alto Networks, SentinelOne, Trend Micro and Wiz are all integrating Claude Opus 4.7 directly into their platforms, while Accenture, BCG, Deloitte, Infosys and PwC are building deployment practices around the product. Schafer described it bluntly: "I think that's basically every name we have an enterprise security right now, all the big consulting firms."
“Cybersecurity is one category where anthropic isn't blocked from the federal market, even with the Pentagon right now”— Jaeden Schafer
Claude Security looks like a constrained sibling to the more open-ended security model Anthropic has been describing internally. The company has talked up a stronger model that is unusually capable at finding and exploiting vulnerabilities, which it has handed to top-tier firms. The public beta is narrower by design — a scanner that produces a list rather than a chat partner that can be coaxed into offensive work, and one that requires explicit access to source code rather than working from the outside in.
For corporate buyers, the appeal is the loop that closes inside Anthropic's own stack: Claude Security identifies the problem, Claude Code fixes it, and the only artefact passed between them is a markdown file. That sidesteps the usual integration tax of plugging an AI tool into existing developer pipelines, and it gives Anthropic a story to tell about agentic workflows that does not depend on building a bespoke API for every customer.
Schafer argued the timing is the under-covered part of the story. Anthropic has been shut out of some Pentagon deals as its tussle with the US defense establishment plays out, but cybersecurity remains a lane the company can still sell into across the federal government. "Cybersecurity is one category where anthropic isn't blocked from the federal market, even with the Pentagon right now, you know, they have this big fight, they're blocked out of some other big deals coming out of the Pentagon."
That makes the launch as much a market-positioning move as a product release. By tying Claude Opus 4.7 into the platforms enterprises already buy from CrowdStrike, Palo Alto and Wiz, and seeding deployment practices at the Big Four consultancies, Anthropic is locking in distribution for a use case where regulators and federal buyers are still willing to write checks. The next test is whether the scheduled scans and patch handoffs hold up in real codebases — and how quickly rivals at OpenAI and Google move to contest the security beat.
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.
