Anthropic has launched Claude Security, a new tool now in public beta for Claude Enterprise customers, designed to help cybersecurity defenders leverage AI to counter increasingly sophisticated AI-powered attacks.
Claude Security scans code for vulnerabilities and suggests patches using Claude Opus 4.7. Unlike traditional pattern-matching tools, the model analyzes how code components interact across files and modules, providing severity, reproducibility, and confidence scores for each finding. The beta also introduces scheduled scans, CSV export, and integrations with Slack and Jira.
Major cybersecurity firms including CrowdStrike, Palo Alto Networks, SentinelOne, and Wiz are integrating Opus 4.7 into their own security products.
The launch comes amid growing concerns about AI's offensive capabilities. Anthropic recently made headlines with its "Mythos" AI model, which the company deemed too dangerous to release due to its ability to autonomously compromise weakly defended networks—a finding confirmed by a British cybersecurity agency. OpenAI's GPT-5.5 has shown similar capabilities. Claude Security aims to give defenders the same AI edge that attackers already have.
Source: Anthropic