What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Anthropic has expanded its Cyber Verification Program (CVP) into three access tiers for verified defenders—but it has not removed cyber safeguards for Claude users generally. The change, announced in October 2026, gives approved organizations more flexibility for sensitive security work while retaining application review, security requirements, monitoring, and the Usage Policy.
What changed in Anthropic’s Cyber Verification Program?
CVP now has three tiers—Defense Access, Red Team Access, and Specialized Access—covering Claude Opus 5.5, Claude Sonnet 5.5, Claude Mythos 5.1, and future models. Anthropic describes the expansion as a way to support defensive work that can be interrupted by safeguards, without making reduced safeguards available to everyone. Anthropic’s announcement and its Help Center overview describe the program and its operating terms.
As an Amazon Associate I earn from qualifying purchases.
| Tier | Purpose and eligibility | Safeguard posture | Access routes |
|---|---|---|---|
| Defense Access | For verified defenders conducting authorized security work. Anthropic’s public materials do not specify one universal eligibility checklist for every applicant. | Some cyber activity remains blocked; this tier is not an unrestricted mode. | First-party Claude.ai, Claude Code, and Anthropic API. Supported third-party enrollment is available for Defense Access, subject to platform conditions. |
| Red Team Access | For verified organizations conducting authorized red-team work. Approval depends on the applicant’s work, verification, and security controls. | More permissive than Defense Access for covered work, but remains subject to program review and policy. | First-party services and supported third-party enrollment. Availability depends on the platform. |
| Specialized Access | Limited to deeply reviewed organizations authorized to test systems whose compromise could endanger lives or disrupt markets, such as flight systems, power grids, telecom networks, interbank transfer infrastructure, or government administrative networks. | Fewest cyber blocks; narrowest intended scope. Anthropic says it reviews each organization in depth with the U.S. government. | First-party services. Third-party platform enrollment does not support Specialized Access. |
The tier descriptions reflect Anthropic’s published program materials; they do not establish a guaranteed approval path or a fixed list of tasks each applicant may perform. Existing Project Glasswing members transition to Specialized Access for current models without reapproval, according to Anthropic.
Free tools Windows power users keep installed
One-click scans. No signup required.
What can Claude users do without CVP?
Anthropic says generally available models remain suitable for ordinary defensive work, including code review, patching known issues, finding vulnerabilities in source code the user owns, and triaging security alerts. More sensitive dual-use activities—including some malware analysis or exploit validation—may still be interrupted by safety classifiers without program access.
#1 Best Overall
CVP is therefore relevant when an organization’s authorized work repeatedly encounters those limits. It is not a general permission to conduct offensive activity, and access does not override Anthropic’s rules: “The Usage Policy still applies in full.” Anthropic may review, narrow, or withdraw a grant. Separately, organizations building client-facing products on these capabilities are governed by Anthropic’s Cyber Productization Policy.
How organizations apply and get access
- Apply through the Verification Portal. Describe the organization’s work and attest to the controls applicable to the requested tier.
- Undergo verification and review. Anthropic says approval depends on the work, its ability to verify the organization, and the organization’s controls; applying does not guarantee approval.
- Use an eligible access route. First-party use includes Claude.ai, Claude Code, and the Anthropic API. Supported cloud and third-party platforms have separate limits. For example, Bedrock access is available only to customers with Enterprise Frontier Safeguards; third-party enrollment supports Defense and Red Team Access, not Specialized Access.
- Maintain the required controls. Requirements vary by tier and deployment and can include phishing-resistant MFA and restrictions on long-lived credentials. Approved organizations remain subject to monitoring and ongoing review.
Organizations already in CVP or Project Glasswing do not need to reapply for the transition described by Anthropic. Data retention is also part of the program’s monitoring approach. Anthropic says eligible organizations may use zero data retention in specified circumstances and describes Enterprise Frontier Safeguards as a forthcoming option for eligible organizations to store data in cloud infrastructure they control; availability and eligibility can change.
Which MFA methods qualify, and when do Defense Access rules apply?
For the applicable CVP requirements, Anthropic lists FIDO2/WebAuthn security keys, passkeys, and smartcards/PIV as phishing-resistant MFA. A security key is one option, not a universal requirement. SMS or voice codes, emailed codes, authenticator-app codes, and push approvals do not qualify under the stated requirements. Anthropic’s detailed security requirements say Defense Access has until December 15, 2026, to adopt phishing-resistant MFA and stop using API keys.
What Anthropic’s published results show—and what they do not
Anthropic describes CyScenarioBench as a test of multi-stage cyber operations under realistic constraints. In its October 2026 evaluation, the company tested Claude Opus 5.5 on 10 challenges, with five attempts per challenge in each tier:
Rank #3
| Condition | Anthropic-reported outcome |
|---|---|
| No CVP | Every task was blocked on the first prompt. |
| Defense Access | 46 of 50 trials were blocked at some point; four succeeded. |
| Red Team Access | No blocks; 34 of 50 tasks succeeded. Anthropic says this was effectively equivalent to the model’s 67.6% success rate without safeguards. |
These are Anthropic’s own benchmark results, not independent validation of real-world outcomes. They illustrate the trade-off the tiers are intended to manage: fewer interruptions can permit more task completion, so access is paired with verification and security controls.
Anthropic also reports vulnerability figures from Project Glasswing and its own scanning. It says partners found at least 129,000 verified software vulnerabilities from April through July 2026, and Anthropic’s open-source scanning found an additional 5,500 from April through October 2026. Through October, Anthropic reported more than 33,000 vulnerabilities rated critical or high. The company says this tally is likely an undercount, based on partial data from 33 partner reports and open-source partnerships; fewer than half of partners had disclosed patched counts, often because fixes remained in progress. These figures are company-reported, not independently audited counts, and they should not be read as the number of vulnerabilities already patched.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →
