Anthropic has expanded its Cyber Verification Program, allowing a wider group of cybersecurity professionals in India to access its most powerful AI models under controlled conditions, aiming to uncover and address critical software vulnerabilities.
- Project Glasswing discovered 129,000+ verified software vulnerabilities early 2026.
- New Cyber Verification Program offers three tiers of AI model access.
- Focus on critical infrastructures like power grids and banking systems.
What happened
Anthropic has expanded its Cyber Verification Program (CVP), integrating previous initiatives like Project Glasswing and the original CVP to give vetted cybersecurity teams greater access to its most advanced AI models. This update introduces three tiers of access, each distinguished by verification requirements and security controls. These tiers include capabilities spanning from incident response and malware analysis to authorized penetration testing and testing of safety-critical systems.
Between April and July, partners in the Glasswing initiative discovered at least 129,000 verified software vulnerabilities. Anthropic’s own open-source scans added another 5,500 vulnerabilities identified through October 2026. The vast majority are rated critical or high severity, and the company believes that the real number of vulnerabilities discovered is likely at least five times higher than recorded, as participation has been limited to a select group of collaborators.
Why it matters
Anthropic’s initiative addresses major security challenges by allowing trusted professionals to test and identify software weaknesses with AI assistance. This is particularly urgent given the growing concern that AI could be used maliciously to exploit software before it is patched or secured. The program balances reduced safeguards for AI models with strict vetting to prevent misuse while amplifying defensive cybersecurity capabilities.
The tiered access approach reflects a careful calibration between openness and risk management, catering to different levels of security sensitivity. From defense-related malware analysis to red team penetration testing, and the highest level focusing on safety-critical infrastructure, the program helps strengthen cybersecurity resilience across multiple sectors critical to India’s digital infrastructure and safety.
What to watch next
Monitor how adoption of Anthropic’s AI tools evolves among India’s cybersecurity teams, especially those operating in critical infrastructure sectors such as power grids, flight systems, and interbank transfers. The effectiveness of this program in uncovering and mitigating vulnerabilities could influence similar initiatives regionally and globally.
The partnership between Anthropic and the US government for member vetting highlights the international collaboration dimension in cybersecurity. Further developments in the Cyber Verification Program’s operational protocols and expansions could impact regulatory and security frameworks, both in India and worldwide, shaping the role of AI in proactive cybersecurity defense.