Home / Technology / AI Now Finds Code Flaws Like a Hacker
AI Now Finds Code Flaws Like a Hacker
21 Feb
Summary
- New AI tool reasons through code like a security researcher.
- It identifies vulnerabilities and suggests software fixes.
- Limited preview access is available for select customers.
Anthropic has introduced Claude Code Security, a new feature for its AI coding tool designed to detect security vulnerabilities within codebases. This advanced capability moves beyond traditional rule-based systems by employing AI that can analyze code with a logic similar to human security researchers.
The tool provides identified issues with severity and confidence ratings, helping security teams prioritize threats. It automatically re-examines findings to reduce false positives. Users will receive identified issues on a dashboard for assessment and action; the AI will not directly alter code.
Claude Code Security is currently in a limited research preview for Enterprise and Team customers. Additionally, maintainers of open-source code repositories can apply for free, expedited access to this new cybersecurity feature.
Anthropic's move follows OpenAI's beta testing of a similar agentic security researcher. The emergence of these AI-driven security tools has reportedly impacted traditional cybersecurity companies, with some seeing stock value declines following the announcement.
