> ## Content Index
> Fetch the complete content index at: https://globalfeed.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Anthropic expands its Cyber Verification Program with three access tiers for security teams
- URL: https://globalfeed.ai/en/anthropic-expands-its-cyber-verification-program-with-three-access-tiers-for-security-teams/
- Published: 2026-10-07T23:54:29.000Z
- Updated: 2026-10-07T23:54:29.000Z
- Description: Anthropic has expanded its Cyber Verification Program, which gives vetted security professionals access to its models with fewer restrictions. The program now has three tiers, Defense, Red Team and Specialized, and each includes Claude Mythos 5.1, Opus 5.5 and Sonnet 5.5.
- Author: GlobalFeed Editor
- Tags: Anthropic, Claude, cybersecurity, penetration testing, x-AnthropicAI, dil-en, elle

**Anthropic** announced on October 6, 2026 that it is expanding its **Cyber Verification Program**. The program lets security professionals whose identity and work have been verified use Claude models for security work with fewer blocks. Publicly available versions of Claude include filters that stop cybersecurity requests that could be misused; for verified teams, those filters are relaxed depending on the type of work.

According to Anthropic's announcement, the expanded program brings two previously separate arrangements under one roof: Project Glasswing, through which critical software organizations used Claude Mythos, and the earlier verification program that reduced safeguards on Opus and Sonnet. Every tier includes Claude Mythos 5.1, Opus 5.5 and Sonnet 5.5, as well as future models.

## Three access tiers

**Defense Access** covers defensive work such as security operations, incident response, malware reverse engineering and vulnerability validation. Security teams defending their own systems, critical infrastructure operators such as regional hospitals or municipal utilities, smaller security firms, open-source maintainers and individual researchers with a record of reported vulnerabilities can apply. Anthropic aims to respond to these applications within a few days.

**Red Team Access** adds authorized penetration testing and red teaming. It is open to in-house and government red teams and to security and penetration testing firms; individual researchers are not eligible. Testing is limited to systems the organization is authorized to test. Actions that could cause physical harm or mass disruption, such as deploying ransomware or damaging physical systems, are still blocked in real time. Review takes a few weeks, and qualifying organizations get Defense Access while they wait.

**Specialized Access** has the fewest blocks. It is open to a limited set of verified organizations authorized to test safety-critical systems such as flight systems, power grids, telecom networks and interbank transfer infrastructure. Anthropic reviews each organization in depth together with the US government. Existing Project Glasswing members move into this tier without reapproval for current models.

## Applying and requirements

Applications go through Anthropic's program page. Applicants must document the security controls their tier requires. Data retention is required so Anthropic can monitor for cyber misuse; for organizations that work without data retention, the company plans to offer an arrangement called Enterprise Frontier Safeguards this fall, which will keep data in cloud infrastructure the organization controls. The program is available on the Claude Platform, Google Cloud Vertex AI and Microsoft Foundry; on Amazon Bedrock it is open only to customers eligible for that arrangement. Existing members keep their settings for earlier models and are automatically evaluated for the new ones.

Users outside the program can keep using the generally available models for code review, patching known issues, finding vulnerabilities in their own source code and triaging security alerts. In the announcement, Anthropic also shares how the tiers differed in its own multi-stage cyber operations test: without program access, tasks were blocked on the first prompt, while Red Team Access ran without blocks.

Anthropic notes that its cyber filters can still block some authorized work and says it will keep refining them. Users who believe their tier should allow a blocked task can file an appeal through the company's form. The published test results and assessments of the program's impact come from Anthropic. For a fuller picture, follow evaluations from independent researchers as well.

Sources: [Anthropic: Expanding the Cyber Verification Program](https://www.anthropic.com/news/cyber-verification-program?ref=globalfeed.ai), [Anthropic: program application page](https://portal.anthropic.com/programs/cvp?ref=globalfeed.ai).