In recent months, there has been growing concern over how cybercriminals, including those linked to state-sponsored activities, have successfully leveraged advanced AI models from the company Anthropic to uncover security gaps in various software systems, including the robust infrastructures of major corporations. In response, Anthropic has announced the launch of similar AI capabilities designed to help its clients identify vulnerabilities.
Anthropic is introducing Claude Code Security, an innovative cybersecurity feature for Claude Code, its renowned AI coding tool. This new feature allows for scanning codebases to detect security vulnerabilities and offers relevant software fixes for human scrutiny.
The distinguishing factor of this tool lies in its methodology. Unlike conventional security solutions that depend on rule-based pattern matching—where code is contrasted against a database of known vulnerabilities—Claude Code Security “analyzes your code akin to a seasoned security investigator,” according to the company.
Vulnerabilities flagged by the tool will be categorized with severity ratings to aid security teams in prioritizing their responses. Additionally, the system will provide “confidence scores” that reflect its certainty about each identified risk, following a secondary analysis to minimize false positives. While the tool does not directly alter any code, it will compile identified issues into a dashboard for the evaluation and action of human security teams.
Currently, Claude Code Security is accessible in a limited research preview for Enterprise and Team customers. Organizations maintaining open-source code repositories can also request expedited, free access.
Anthropic is not the only firm introducing such features; OpenAI commenced beta testing Aardvark, an AI-driven security researcher utilizing GPT-5, last October.
Recommended by Our Editors
There are speculations that traditional cybersecurity companies may face challenges due to the emergence of tools like Anthropic’s latest offering. According to SiliconANGLE, CrowdStrike Holdings experienced a nearly 8% decline in stock value immediately following the announcement, while Cloudflare also dropped over 8%.
In a November 2025 interview with PCMag, Boris Cherny, Head of Claude Code, offered insights into future developments for the application. He suggested that upcoming models “will operate for extended periods without human oversight,” and hinted at enhanced integrations with other AI systems.
Get Our Best Stories!
Your Daily Dose of Our Top Tech News
Thanks for signing up!
Your subscription has been confirmed. Keep an eye on your inbox!
About Our Expert
Experience
I’m a reporter who covers weekend news. Before joining PCMag in 2024, I contributed to various platforms including BBC News, The Guardian, The Times of London, The Daily Beast, Vice, Slate, Fast Company, The Evening Standard, The i, TechRadar, and Decrypt Media.
As a long-time PC gamer, I have witnessed the evolution of technology and its impact on human experiences. My reporting spans a wide range of topics, from cryptocurrency scandals to the art world, conspiracy theories, UK politics, and international relations.
