Anthropic Expands Cyber Verification Program, Opens Restricted AI Access to More Security Groups

Anthropic's expansion of its Cyber Verification Program (CVP) now grants restricted AI access to additional security groups, offering advanced cyber capabilities and reduced blocking for vetted participants. For website owners, this means a broader pool of sophisticated AI agents—potentially equipped with enhanced penetration‑testing and reconnaissance tools—will have legitimate pathways to interact with site infrastructure. Those enrolled in the CVP can probe for vulnerabilities more aggressively while evading typical bot detection filters, increasing the risk of unnoticed security assessments or malicious exploitation under the guise of “verified” traffic. To mitigate exposure, first update your llms.txt file to explicitly list Anthropic’s CVP endpoints (e.g., https://api.anthropic.com/cvp/*) and declare your preferred interaction policies, such as “no automated vulnerability scanning.” Second, integrate a dedicated CVP‑aware bot‑tracking rule in your analytics platform that flags any requests bearing Anthropic’s verification tokens or user‑agent signatures, allowing you to audit and, if necessary, block unexpected activity. Finally, schedule quarterly reviews of your security logs for CVP‑related traffic spikes and coordinate with Anthropic’s compliance team to ensure any legitimate testing is coordinated and documented.