Anthropic is expanding its Cyber Verification Program (CVP) to give more vetted cybersecurity professionals and organizations access to its most capable AI models, according to an October 6 Reuters report.

The program uses three tiers with different eligibility criteria and usage restrictions based on the intended use case and risk level.

The expansion gives approved teams access to models with fewer safeguards for activities such as incident response, authorized penetration testing and research involving safety‑critical systems.

Anthropic said the program is designed to give security teams greater access while maintaining controls around how the models are used.

AI Access Requires Risk-Based Governance

For chief data officers (CDOs), the program highlights the need to govern access to increasingly capable AI systems based on their intended use.

Organizations need to determine which users and teams should have access to advanced models, what data those models can access and what controls should apply to each use case.

Anthropic’s three‑tier structure provides different levels of access and verification:

• Its Defense tier supports incident response, malware analysis and related work by security teams, critical infrastructure operators, open‑source maintainers and researchers with a track record.
• The Red Team tier adds authorized penetration testing and red‑teaming, and is limited to organizations.
• The Specialized tier is for a small, highly vetted group authorized to test safety‑critical systems such as power grids, flight systems and interbank transfers; members are vetted with the U.S. government, and existing Project Glasswing participants move into this tier.

The approach also highlights the importance of monitoring AI access after approval.

As models interact with sensitive systems and data, AI governance teams need visibility into who is using them, what information they can access and whether their activity remains within approved boundaries.

This makes AI governance roles and accountability increasingly important as organizations expand access to advanced AI systems.