Claude Code Security scans code for vulnerabilities and suggests fixes
Anthropic introduces Claude Code Security, a new AI tool that detects and resolves vulnerabilities in codebases. The tool itself does not make changes to the code; it only provides suggestions.
More than traditional tools
Traditional analysis tools primarily look for known patterns (such as poor encryption), but Claude analyzes how code actually works: data flows, component interactions, and logical dependencies. According to Anthropic, this is much more effective.
Every vulnerability undergoes a verification process to minimize errors. The tool assigns a severity and reliability score to each issue, and findings appear in a dashboard where developers can review and approve suggested patches with the click of a button.
The solution runs on Claude Opus 4.6, the latest version of the Claude model.
Research preview
According to Anthropic, internal tests and hackathons identified more than 500 previously undiscovered vulnerabilities in open-source projects. Some had been present for decades. Claude Code Security is currently available in a limited research preview. Enterprise and Team customers can request early access, while open-source users can also sign up for an accelerated testing phase.
Falling stocks
Following the announcement, a striking number of competing stocks fell sharply, reports SiliconANGLE. CrowdStrike Holdings Inc. recorded a loss of nearly 8 percent, and Cloudflare Inc. also saw a decline of just over eight percent. This is the second time this month that Anthropic has shaken up the stock market.