Anthropic is giving more cybersecurity teams access to its latest Claude models with fewer limits. The company’s new Cyber Verification Program will offer Claude Opus 5.5, Claude Sonnet 5.5 and Claude Mythos 5.1 to approved users. Previously, the company launched its Project Glasswing which was aimed at securing world’s most critical software. Notably, this initiative uncovered more than 100,000 software vulnerabilities between April and July 2026. Anthropic's own open-source scanning found 5,500 more between April and October.
Thus, once again the company is planning a revamped Cyber Verification Program, or CVP where the AI giant will run two programs. The first will be Glasswing, which will offer verified companies the access to Claude Mythos, Anthropic's most cyber-capable family of models.
The next is the original CVP where chosen security teams will receive reduced safeguards on Claude Opus and Sonnet models. Users will be placed into three levels based on the type of work they do and how much access they need.
The program has three levels called Defense, Red Team and Specialized. Defense is meant for work such as dealing with security incidents and studying malware, Red Team is for approved security tests, and Specialized has the fewest limits and is meant for selected groups working on important systems, including power, aviation and financial networks.
Anthropic says users will be checked before they receive access. The idea is to give powerful AI tools to people who can use them for security work without making them easy to misuse.
Also Read: Australia Weighs New AI Breach Rules as OpenAI, Anthropic Show Support
Fable 5.1 and Mythos 5.1 are based on the same model. The main difference is in their safety limits. Fable has tighter safeguards and is available to more users. Mythos has fewer limits for trusted users working in fields such as cybersecurity and life sciences.
This could help researchers work on real security problems without the model stopping them too early. A security team may need to inspect risky code or test a weakness in a system. Mythos is designed to give approved users more room to do that work.
Anthropic is opening access to its stronger Claude models, but it is not giving the same level of access to everyone. The company wants trusted security teams to find software flaws faster while keeping these powerful tools under tighter control.