Anthropic's Claude Security exits limited preview to reach all Enterprise customers, offering AI-driven code vulnerability detection and remediation as the market race intensifies.
Anthropic moved Claude Security into public beta on Thursday, opening access to all Claude Enterprise subscribers. The product scans code for vulnerabilities and generates fixes, graduating from a limited research preview that the company says reached hundreds of organizations across sectors.
According to CRN, the offering was previously called Claude Code Security and first announced in late February. It runs on existing Claude infrastructure and does not use Claude Mythos, Anthropic's unreleased next-generation model. Claude Team and Max subscribers are excluded from the beta for now; Anthropic says access for those tiers is "coming soon," without specifying a date.
The Mythos backdrop
The launch is timed against a disclosure that rattled the security industry. On April 7, Anthropic revealed that Claude Mythos Preview demonstrates coding ability that can surpass "all but the most skilled humans." That finding sharpened the threat model for enterprise software teams: an artificial intelligence system capable of writing expert-level code can, by the same logic, read and exploit software vulnerabilities at industrial scale.
In Thursday's announcement, Anthropic named that risk explicitly. The company stated that current models are already effective at locating flaws in software and that the next generation will be capable of autonomously exploiting those same flaws. That is unusually candid framing from a vendor launching a product to do what attackers could do first.
The enterprise race
This launch sits inside a broader competition for enterprise developer and security tooling. As CNET noted this week, the AI industry has pivoted sharply toward enterprise work tools, with products like Claude Code competing alongside OpenAI's Codex for engineering budgets. OpenAI closed a $122 billion funding round in late April at an $852 billion post-money valuation, and separately plans to nearly double its workforce to 8,000 by year-end, with most new roles targeting product, engineering, and sales.
Security tooling is a particularly valuable category to enter early. Tools embedded in CI/CD pipelines and developer workflows are replaced infrequently and at high switching cost. Anthropic is betting that capturing those workflows now, while the market is still forming, creates durable enterprise revenue leverage.
What evaluators should ask
Anthropic has not published benchmarks comparing Claude Security's detection rates against established static-analysis platforms or AI-native competitors. The hundreds of organizations in the limited preview signal demand, but say nothing about false-positive rates, coverage depth, or integration fidelity with existing SAST and DAST toolchains.
Those gaps are expected at the beta stage, but they are precisely the questions that will drive or stall procurement decisions over the next several quarters.
Anthropic's stated logic is stark: if the next generation of AI can autonomously exploit code, the window for deploying defensive tools is narrowing. Whether Claude Security builds a durable position before better-capitalized competitors close the gap depends on how quickly the company puts rigorous performance data in front of security teams.
Frequently asked questions
What is Claude Security?
Claude Security is Anthropic's AI-powered tool for finding vulnerabilities in code and generating suggested fixes. It entered public beta on April 30, 2026, and is currently available to Claude Enterprise subscribers.
Who can access the Claude Security public beta?
All Claude Enterprise customers have access as of Thursday. Claude Team and Max subscribers are not yet included; Anthropic says access for those tiers is coming soon, without specifying a date.
Does Claude Security use the Claude Mythos model?
No. Anthropic confirmed the beta does not use Claude Mythos Preview, its unreleased next-generation model. The tool runs on existing Claude infrastructure.
How does it compare to traditional code security tools?
Claude Security is positioned as an AI-native alternative to conventional static-analysis tools. Independent benchmarks comparing its detection rates to established SAST or DAST platforms are not yet publicly available.