Anthropic’s new Claude Security tool is now available in public beta to Enterprise customers, giving them the ability to scan their codebases for potential vulnerabilities and generate patches.
Operating under the Claude Opus 4.7 model, it was originally released earlier this year under the name Claude Code Security and has already been extensively tested by “hundreds of organizations” of various sizes, according to Anthropic. Their feedback has apparently shaped the release.
The release marks another significant stage in Anthropic’s ongoing cybersecurity efforts following the recent headlines surrounding Claude Mythos, a model considered so powerful that the generative AI vendor is only making it available to selected partners to fix critical flaws rather than offering it publicly, as part of its Project Glasswing program.
Although not in the same league as Mythos, Claude Security will be accessible to all Enterprise customers using Opus 4.7, which the vendor says is among the strongest models available for discovering vulnerabilities and complex issues that might otherwise be missed.
“It comes with scheduled and targeted scans, easier integration with audit systems, and improved tracking of triaged findings. No API integration or custom agent build is required: if your organization uses Claude, you can start scanning today,” according to the Anthropic blog post about the release.
In addition, Opus 4.7’s newly enhanced security capabilities are being integrated into the software that many enterprise customers already use. Tech partners such as Microsoft Security, CrowdStrike and TrendAI are embedding Opus 4.7 into their existing tools, while services partners such as Accenture, Deloitte, PwC and Infosys are helping organizations deploy Claude-integrated solutions, Anthropic said.
Once a scan is started, the tool does not search specifically for known vulnerabilities; instead, it analyzes how components interact across modules and files, traces data flows and reads source code. It then provides an explanation of its findings, including the level of threat and instructions for a targeted patch, which users can open in Claude Code on the web.
Among the feedback from early testers, which has guided the release, is that the need for high-quality detection is paramount; that false positives are a problem and quick fixes are essential; and that teams prefer ongoing coverage rather than isolated audits.
One of the early testers was DoorDash, whose vice president and chief security officer, Suha Can, said in a statement: “Claude Security surfaces deep vulnerabilities accurately, and pipes findings right into our workflows so engineers can act on them in context.”
While the tool is available to Claude Enterprise clients immediately, Team and Max customers will have access soon.
