Anthropic has introduced Claude Security in public beta, an AI-driven solution designed to assist cybersecurity teams in identifying vulnerabilities across codebases and automatically producing patches, advancing the company’s focus on software security.
- AI model Opus 4.7 powers deep code vulnerability detection and patch generation
- New features include scheduled scans, triage note-taking, and export options
- Integrated with leading cybersecurity platforms like CrowdStrike and Palo Alto Networks
What happened
Anthropic PBC announced the public beta release of Claude Security, a tool designed to help cybersecurity teams identify vulnerabilities and generate fixes directly within their codebases. This tool is part of the broader Claude Enterprise subscription service aimed at large organizations, evolving from its initial research preview launched earlier this year. Since the preview stage, hundreds of organizations have used Claude Security to uncover and address security flaws that existing solutions overlooked for years.
Powered by Anthropic’s Opus 4.7 AI model, Claude Security functions by reasoning through code similarly to a human cybersecurity researcher. It analyzes entire codebases, tracing data flows, examining interactions among code components, and provides vulnerability assessments with confidence ratings. This allows users to understand the severity and exploitability risk of detected issues while facilitating immediate fixes within the code editing environment.
Why it matters
Claude Security’s ability to inspect whole codebases and infer security risks beyond known vulnerability patterns represents a significant advancement in automated security analysis. By offering detailed explanations and confidence levels along with patch suggestions, it helps security analysts and developers accelerate remediation processes, reducing the traditional lag time between detection and fix deployment.
The product’s integration potential with major cybersecurity firms such as CrowdStrike, Palo Alto Networks, SentinelOne, Trend Micro, and Wiz enhances its practical impact in the industry. These partnerships bring Claude’s AI-powered insights into established security platforms, potentially increasing enterprise-wide threat detection and response capabilities globally.
What to watch next
Following initial user feedback, Anthropic has added functionality including scheduled vulnerability scans for continuous security coverage and options to dismiss findings with documented reasons, improving auditability and alignment with organizational security workflows. Users can also export findings in standard formats compatible with existing audit and compliance tools.
Looking ahead, the industry will be monitoring how Claude Security’s AI reasoning approach compares with other emerging automated security systems in real-world deployments. Additionally, the impact of Anthropic’s broader Project Glasswing initiative, which leverages different AI models for security, and expanding collaborations with technology partners will shape the tool’s adoption and evolution in next-generation cybersecurity frameworks.