Anthropic’s Cyber Mission and Cyber Verification Program, explained
Anthropic opened its most capable Claude models to more security defenders in October 2026. On Oct. 6, it expanded its Cyber Verification Program (CVP) into three tiers of access with fewer safety blocks for vetted teams. On Oct. 8, it launched the Anthropic Cyber Mission, which it called "a long-term commitment to securing the systems everyone depends on."
What is the Cyber Verification Program, in one sentence?
The Cyber Verification Program gives security professionals Anthropic has vetted access to Claude with fewer of the automatic blocks that normally stop cyberattack-related requests.
Why does Claude block security work at all?
Anthropic calls cybersecurity "inherently dual use": the skills that find and fix a software flaw can also exploit it. Generally available Claude models block most cyber work, though they can still review code, patch known issues, find flaws in code you own and triage security alerts, according to the Oct. 6 post.
How does the program work?
The expanded program merges two earlier efforts: Project Glasswing, which gave critical-software organizations access to Claude Mythos, and the original CVP, which reduced safeguards on Claude Opus and Sonnet. Every tier includes Claude Opus 5.5, Claude Sonnet 5.5, Claude Mythos 5.1 and future models.
- Defense Access is for defensive work such as incident response, malware analysis and vulnerability validation. It is open to companies, nonprofits, universities, governments, open-source maintainers and individual researchers with a track record of reported vulnerabilities. Anthropic aims to respond "within a few days."
- Red Team Access adds authorized penetration testing for in-house, government and commercial red teams, on systems they are authorized to test. Individuals are not eligible, and reviews may take "a few weeks."
- Specialized Access has the fewest blocks and is reserved for a limited set of organizations that test systems affecting lives or markets, such as flight systems, power grids and interbank transfers. Anthropic reviews each one in depth with the U.S. government.
Applying takes these steps, according to Anthropic:
How to join Anthropic's Cyber Verification Program
- Submit an application through the CVP portal and choose a tier.
- Provide verification and proof of the security controls your tier requires.
- Defense Access aims for a reply within a few days; Red Team Access may take a few weeks.
- Workspace admins assign the program to specific workspaces in the Claude Console.
How did Anthropic test the tiers?
Anthropic ran Claude Opus 5.5 on 10 challenges from a benchmark called CyScenarioBench, five attempts each, under each tier. Without CVP access, every task was blocked at the first prompt. With Defense Access, 46 of 50 attempts were blocked at some point. With Red Team Access, none was blocked and 34 of 50 were completed, the same 67.6% success rate as with no safeguards.
What did the earlier program find?
Glasswing partners found at least 129,000 verified software vulnerabilities between April and July 2026, and Anthropic's open-source scanning found 5,500 more between April and October 2026, the company says. More than 33,000 of them are rated critical or high severity. Anthropic calls these figures a lower bound, since fewer than half of its partners reported patch numbers.
What is the Cyber Mission?
The Cyber Mission starts in two areas. Its Critical Infrastructure Defense Program brings Claude models, on-site engineers and threat research to companies that protect power grids, water systems and transportation networks; founding partners include CrowdStrike, Dragos, Palo Alto Networks and Rockwell Automation. Its OSS Scanner is a free, opt-in service that periodically scans open-source projects with Anthropic's most capable models. Its reports are generated by models and sent without human review, and Anthropic expects more than 90% of them to be correct.
What doesn't Anthropic say?
Anthropic gives no dollar amounts for the funding it says it has given to the Python Software Foundation, OpenSSF, the Apache Software Foundation and others. The vulnerability counts are self-reported, and the tier test covers one model on one benchmark. Anthropic also notes that enrolled organizations must allow data retention so it can monitor for misuse.
Context: why now?
Anthropic says it expects AI to favor defenders within two years, but warns the near term may be different: exploiting flaws has become cheaper, while verifying and fixing them is still slow.
FAQ
What is Anthropic's Cyber Verification Program?
A program that gives vetted security professionals access to Claude models with fewer cyber safety blocks, in three tiers: Defense, Red Team and Specialized Access.
Can an individual security researcher join?
Yes, for Defense Access, if they have a track record of reported vulnerabilities. Red Team Access is not open to individuals.
What does Claude still block for verified users?
Even with Red Team Access, Claude blocks in real time actions that could cause physical harm or mass disruption, such as deploying ransomware or damaging physical systems.
What is Anthropic's OSS Scanner?
A free, opt-in service from Anthropic's Cyber Mission that periodically scans enrolled open-source projects for vulnerabilities and sends model-generated reports with a proof of concept and a suggested fix.
Sources
Cite this
Anthropic’s Cyber Mission and Cyber Verification Program, explained, New Mind Museum, November 6, 2026. https://newmindmuseum.com/safety/2026/11/anthropic-cyber-mission-verification-program-explained/
New Mind Museum. (2026, November 6). Anthropic’s Cyber Mission and Cyber Verification Program, explained. https://newmindmuseum.com/safety/2026/11/anthropic-cyber-mission-verification-program-explained/
“Anthropic’s Cyber Mission and Cyber Verification Program, explained.” New Mind Museum, 6 Nov. 2026, https://newmindmuseum.com/safety/2026/11/anthropic-cyber-mission-verification-program-explained/.
@misc{new-mind-museum-2026-anthropics-cyber-mission-and-cyber-verif,
author = {{New Mind Museum}},
title = {Anthropic’s Cyber Mission and Cyber Verification Program, explained},
year = {2026},
url = {https://newmindmuseum.com/safety/2026/11/anthropic-cyber-mission-verification-program-explained/}
}Free to quote and cite with a link to this page.