Anthropic's Claude Security Enters Enterprise Beta After Catching Years-Old Vulnerabilities

Anthropic today announced that Claude Security, its AI-powered code vulnerability scanner, is now available in beta for Claude Enterprise customers after a successful closed preview. The tool, integrated into Claude Code on the web, scans entire codebases for security flaws and recommends fixes. Support for Team and Max plan users is expected soon.

“Hundreds of organizations have already used Claude Security to fix issues in their production code that existing tools had missed for years,” an Anthropic representative said.

Background

Launched in private preview in February, Claude Security was developed alongside Anthropic’s broader security initiatives, including Project Glasswing. The tool uses multiple parallel AI agents to analyze source code and data flows, building a comprehensive attack surface map rather than relying on known vulnerability patterns.

Powered by the Opus 4.7 large language model, Claude Security balances detection accuracy with a reduced risk of false positives. While not as advanced as the still-private Mythos model, Opus 4.7 enables the tool to uncover subtle flaws that conventional scanners overlook.

How It Works

During a scan, multiple agents examine the codebase simultaneously. When a potential issue is detected, a secondary validation pipeline automatically verifies the finding before an alert is triggered. “Claude challenges its own findings to ensure fewer false positives,” the spokesperson explained.

Every detection includes a recommended patch that security teams can review and approve directly within Claude Code. This tight integration reduces the turnaround time between discovery and remediation. “Users can open a Claude Code session to work through the patch in context, instead of days of back-and-forth between security and engineering,” the company stated.

New Features in Beta

During the preview period, Anthropic added several capabilities based on user feedback:

• Scheduled scans – run recurring security checks automatically to maintain continuous coverage.
• Dismiss with comments – annotate findings to facilitate team collaboration and audit trails.
• CSV and Markdown exports – integrate scan results into existing security dashboards and reporting tools.

What This Means

Claude Security bridges the gap between vulnerability detection and remediation. By pairing scanning with patch generation in Claude Code, organizations can move from identification to fix in a single workflow, eliminating the traditional back-and-forth between security and engineering teams.

For enterprises, this means faster response times to emerging threats and reduced exposure to exploits. The tool also complements Anthropic’s separate Code Review product, which analyzes pull requests. Together, they form a layered defense for development pipelines.

The beta rollout signals Anthropic’s deepening commitment to developer-first security solutions—a space where speed, accuracy, and context are critical. As more organizations adopt AI-assisted coding, proactive vulnerability scanning becomes essential infrastructure.

Back to Background | How It Works | New Features | What This Means