Anthropic expands Cyber Verification Program with new access tiers

Comcast: Comcast is a major telecommunications company that collaborated with Anthropic through Project Glasswing on cybersecurity initiatives. Its partnership illustrates practical applications of the models in protecting large-scale infrastructure. Anthropic: Anthropic develops advanced AI systems with an emphasis on safety and reliability, primarily through its Claude family of large language models. The company is announcing the expansion of its Cyber Verification Program to provide verified security professionals with tiered access to its most capable models for both defensive cybersecurity and authorized offensive tasks such as penetration testing. Booz Allen: Booz Allen is a professional services firm that participated in Anthropic's Project Glasswing to enhance vulnerability identification in client systems. Its experience is referenced as an example of how advanced models can accelerate defensive cybersecurity outcomes. Google Cloud: Google Cloud provides enterprise cloud infrastructure and AI platform services. The Cyber Verification Program is available to eligible users through Google Cloud's Vertex AI offering. Amazon Bedrock: Amazon Bedrock is a managed service for deploying foundation models on AWS infrastructure. The Cyber Verification Program is available on Amazon Bedrock only for customers meeting Enterprise Frontier Safeguards criteria. Claude Opus 5.5: Claude Opus 5.5 is one of Anthropic's flagship AI models optimized for complex reasoning and task execution. It is being offered through the Cyber Verification Program's tiers to enable security professionals to perform defensive operations, red-teaming, and specialized testing under controlled conditions. Claude Mythos 5.1: Claude Mythos 5.1 is an Anthropic AI model previously accessible through select programs like Project Glasswing. It is now integrated into the Cyber Verification Program across multiple access tiers for use by qualifying security organizations. Claude Sonnet 5.5: Claude Sonnet 5.5 is an Anthropic AI model designed for high performance across a range of applications. It is included in the expanded Cyber Verification Program to support verified security professionals in tasks such as incident response and vulnerability analysis. Microsoft Foundry: Microsoft Foundry delivers enterprise AI solutions and platform capabilities. The Cyber Verification Program can be accessed by qualified organizations via Microsoft Foundry. Project Glasswing: Project Glasswing was an Anthropic initiative that provided select organizations securing critical software with access to advanced Claude models for vulnerability discovery. It is now being folded into the broader Cyber Verification Program to scale defensive capabilities across more security teams. Cyber Verification Program: The Cyber Verification Program is Anthropic's framework for granting security professionals controlled access to AI models with reduced safeguards tailored for cyber work. The expanded version integrates prior initiatives and introduces three access tiers—Defense Access, Red Team Access, and Specialized Access—each with specific eligibility requirements and security controls to balance capability with misuse prevention. Tiered Access Model: The expanded program introduces differentiated verification and control levels that match the scope of defensive, red-teaming, or specialized safety-critical work. Platform Availability: The Cyber Verification Program operates across multiple major cloud environments to support enterprise and government security teams. Program Consolidation: Anthropic is merging its Project Glasswing and prior Cyber Verification Program efforts into a unified, tiered offering to broaden access for security organizations.