Artificial Intelligence & Machine Learning
,
Next-Generation Technologies & Secure Development
,
The Future of AI & Cybersecurity
Flaw Finding Model Integrated into a Slew of Cybersecurity Platforms

Claude artificial intelligence maker Anthropic announced Thursday wider availability of a model it described as its second-most powerful model for finding and patching software flaws.
See Also: Agentic AI and the Future of Automated Threats
Claude Security, running on the Opus 4.7 model, can provide a detailed explanation of a vulnerability, rank its confidence over how real and severe the flaw is, as well as its likely impact, and how it can be reproduced, Anthropic wrote in a blog post.
“It also generates instructions for a targeted patch, which users can open in Claude Code on the web to work through the fix in context.”
Anthropic is making Claude Security available as a “public beta” for enterprise customers, with guardrails enacted for users who haven’t verified themselves as cybersecurity professionals.
Mythos Preview, the model that Anthropic rolled on April 7 and claimed was too dangerous to release to the public, remains restricted to roughly 50 companies allowed to sign onto Anthropic’s Project Glasswing. The Wall Street Journal Thursday reported that Anthropic has proposed letting roughly another 70 companies have access to Mythos but is opposed by the White House. The Trump administration is involved in the Mythos rollout due to the model’s national security risks.
Anthropic also said Thursday that technology partners, including CrowdStrike, Microsoft Security, Palo Alto Networks, SentinelOne, TrendAI, and Wiz are embedding Opus 4.7 into their products.
“By integrating one of the world’s most advanced AI models, we are empowering our customers to outpace automated threats,” said Palo Alto Networks.
Claude Security first saw distribution in February as a research preview named Claude Code Security (Why Claude Code Security Has Shaken the Cybersecurity Market).
Lessons drawn from two months of use informed this new iteration, Anthropic said. Changes include a multi-stage validation of findings meant to drive down false positives and an option to schedule scans. Users can also now scan a particular directory, dismiss findings, export findings and send scan results to project management tools.
Claude’s entry into the vulnerability scanning arena has repeatedly sent jolts through the market but the weeks and months since the product announcements have allowed for notes of skepticism to well up. “Most CVEs never get ‘weaponized’ for in-the-wild campaigns, and for reasons beyond technical difficulty,” wrote cybersecurity entrepreneur Jeremiah Grossman in a Linkedin post. “More capability doesn’t change what attackers need. It just makes it easier to produce things they still won’t use. The common mistake is to treat ‘might work,’ or ‘works in the lab as ‘will be used.’ That leap is where a lot of the current confusion in cybersecurity starts.”
“This is a good post,” wrote cybersecurity maven Kevin Beaumont, who decried Gen AI as “people panicking about stuff they do not understand and have no real world context about.”
