Details
- Anthropic is expanding its Cyber Verification Program to give verified security professionals broader access to advanced Claude models.
- Eligible participants can access Claude Mythos 5.1, Claude Opus 5.5, and Claude Sonnet 5.5 under safeguards designed for defensive cybersecurity work.
- The program is structured around tiered access, allowing organizations to receive capabilities matched to their security work and controls.
- Approved uses include security operations, incident response, malware analysis, reverse engineering, vulnerability validation, and other authorized defensive tasks.
- Anthropic’s approach reduces some cyber-related blocking while retaining applicant verification and security requirements.
- The expansion follows Anthropic’s earlier restricted-access programs for its most cyber-capable models and broadens availability beyond a smaller group of trusted organizations.
Impact
The expansion reflects a broader industry shift toward controlled access rather than unrestricted release of high-capability cyber models. It gives vetted defenders more useful tooling while preserving safeguards against misuse. Anthropic is also positioning the program alongside comparable trusted-access efforts from major rivals, including OpenAI’s cyber-defense initiatives, making verification, monitoring, and tiered permissions increasingly central to competition in AI security products.
