AI Development

Anthropic’s Cyber Verification Program: what Defense Access unlocks in Claude and Mythos 5.2

Defense Access, the defensive tier of Anthropic’s Cyber Verification Program, eases Claude’s blocks on malware analysis, vulnerabilities and incident response.

Luca Di Domenico

Luca Di Domenico

· 4 min read

In short

Anthropic approved me for Defense Access, the defensive tier of its Cyber Verification Program, on November 9, 2026, the day Claude Mythos 5.2 launched. Defense Access eases Claude’s blocks on vulnerability triage, malware analysis, reverse engineering, detection rules and incident response, and it includes Mythos 5.2, Opus 5.5 and Sonnet 5.5. In Anthropic’s CyScenarioBench test it still blocks 46 of 50 offensive operations.

As of this week, I can use Claude for cybersecurity work that a regular account often gets blocked on. Anthropic approved my application for Defense Access, the defensive tier of its Cyber Verification Program. The approval arrived the day Anthropic launched Claude Mythos 5.2, on November 9, 2026.

What is Anthropic’s Cyber Verification Program?

The Cyber Verification Program is restricted access to Claude models for verified security professionals. Members get fewer interruptions from the cyber safeguards Anthropic applies to its most capable models. The program is not available on the Free plan, and Anthropic’s Usage Policy still applies in full.

On October 6, 2026, Anthropic restructured the program into three tiers:

  • Defense Access: defensive work. Security operations, incident response, malware reverse engineering, vulnerability analysis and validation.
  • Red Team Access: adds authorized penetration testing and red teaming. Open to organizations only, and only on systems they are authorized to test.
  • Specialized Access: the tier with the fewest blocks, for a small set of organizations authorized to test critical systems such as power grids and telecom networks. Anthropic reviews every application together with the US government.

What does Defense Access unlock?

Defense Access is built for people who defend systems they own or maintain. In the approval email, Anthropic lists these areas:

  1. vulnerability triage and validation;
  2. malware analysis;
  3. reverse engineering;
  4. detection engineering, meaning the rules that recognize an attack in progress;
  5. incident response.

On a regular Claude account, these tasks often get blocked. The same analysis that defends a system can be used to attack it, so Claude’s filters tend to stop it. The Cyber Verification Program moves the check upstream: Anthropic verifies who is asking and, in exchange, relaxes the filters on defensive work.

Anthropic measured the difference on CyScenarioBench, a benchmark of multi-stage cyber operations, using Claude Opus 5.5 and 50 trials per tier:

  • without the program: every task blocked on the first prompt;
  • with Defense Access: 46 of 50 trials blocked at some point, 4 completed;
  • with Red Team Access: no blocks, 34 of 50 tasks completed.

So Defense Access opens up defensive work and still stops almost every offensive operation. Authorized attack work requires the Red Team tier.

Does Defense Access include Claude Mythos 5.2?

Yes. According to the Claude Help Center, every tier of the Cyber Verification Program includes Mythos, Opus and Sonnet models, starting with Claude Mythos 5.2, Mythos 5.1, Opus 5.5 and Sonnet 5.5. The program adjusts the cyber safeguards on these models by tier.

Claude Mythos 5.2 came out on November 9, 2026. It is the same model as Claude Fable 5.2 without certain cybersecurity and biology safeguards, and only organizations verified through Anthropic’s programs can use it. Anthropic calls it its most capable model for cybersecurity engineering: on the internal CVE Rediscovery evaluation it found 10.2% more vulnerabilities than Opus 5.5 and 14.4% more than Mythos 5.1.

Who can apply for Defense Access?

Defense Access is the only tier open to individual professionals as well as organizations. The Help Center mentions independent researchers, open-source maintainers and bug bounty hunters. Anthropic aims to answer Defense Access applications within a few days.

The main requirements:

  • a paid plan: Pro, Max, Ultra, Team, Enterprise or API;
  • data retention, which Anthropic requires from every member to monitor misuse;
  • for organizations, multi-factor authentication and API keys replaced at least every seven days;
  • from December 15, 2026, phishing-resistant multi-factor authentication.

After approval there is one more step. The organization’s admin has to assign the program to the right roles and workspaces. On Team plans, access applies to all users automatically.

What can you do with this access?

Over the next few weeks I will show what Defense Access makes possible in practice. If you want to start with a regular account, there are already useful tools: I wrote about running a security audit with Claude Code and Cloudflare’s skill.

If you defend systems you own or maintain and Claude stops right on the analysis you need, the Cyber Verification Program is the official way to unlock it.

Sources

  • Expanding the Cyber Verification Program — Anthropic announcement, October 6, 2026: the Defense, Red Team and Specialized Access tiers, eligibility, Defense Access answers within a few days, and CyScenarioBench results with Opus 5.5 (everything blocked without the program, 46 of 50 blocked with Defense Access, no blocks and 34 of 50 tasks completed with Red Team Access)
  • Cyber Verification Program | Anthropic Help Center — Page updated October 6, 2026: Defense Access open to organizations and individuals, Pro, Max, Ultra, Team, Enterprise and API plans, mandatory data retention, MFA and API keys rotated every 7 days, phishing-resistant MFA from December 15, 2026, models included starting with Mythos 5.2, Mythos 5.1, Opus 5.5 and Sonnet 5.5
  • Introducing Claude Fable 5.2 and Claude Mythos 5.2 — Anthropic announcement, November 9, 2026: Mythos 5.2 is the same model as Fable 5.2 without certain cybersecurity and biology safeguards, reserved for verified organizations; on CVE Rediscovery it finds 10.2% more vulnerabilities than Opus 5.5 and 14.4% more than Mythos 5.1
  • Claude Mythos 5.2 - Claude Platform Docs — Documentation checked November 11, 2026: released November 9, 2026, available only to organizations verified through Anthropic’s programs such as the Cyber Verification Program

Want to know what this means for your product?

If you're building software with AI and want a second opinion on how you're working, tell me about your project.

Tell me about your project
CybersecurityClaudeAnthropicAI Agents

Follow the blog

One email for every new article

I only write when an article is out: title, summary and link. You can unsubscribe with one click from any email.

RSS feed

An RSS feed is the up-to-date list of articles: add it to a reader app like Feedly or Inoreader and new articles show up there, no email needed.

lucadidomenico.studio/en/blog/rss.xml