Claude Security brings AI-powered vulnerability detection to production codebases with the Opus 4.7 model, offering scheduled scans, validation to reduce false positives, and auto-generated patches.
Anthropic has opened Claude Security to public beta for Claude Enterprise customers, bringing AI-powered vulnerability detection directly into production codebases without requiring custom tooling or API integrations. Claude Security leverages the Opus 4.7 model to perform end-to-end security analysis, scanning for vulnerabilities, validating each finding to reduce false positives, and generating suggested patches for developer review before deployment.
Claude Security first appeared as a research preview in February 2026, and since then hundreds of organizations have run it against production code, surfacing vulnerabilities that existing scanners had missed. New capabilities added based on early adopter feedback include scheduled scans for recurring security checks, directory-level targeting to focus on specific paths, CSV and Markdown export formats, webhook notifications for real-time alerts, and persistent dismissals that carry forward across scans to reduce noise over time.
The addition of validation logic to cut false positives is particularly notable. One of the biggest pain points with automated scanners is the volume of noise they generate, leading security teams to deprioritize findings or ignore alerts altogether. By pairing detection with model-driven validation, Claude Security aims to deliver a higher signal-to-noise ratio than traditional static analysis tools. Anthropic noted the tool is designed as a direct on-ramp with no agent builds or API wiring required.
For enterprise security teams looking to scale vulnerability coverage without expanding headcount or building internal AI infrastructure, Claude Security represents a low-barrier entry point.
Source: Cyber Security News — Anthropic Launches Claude Security in Public Beta for Enterprise Custo
