The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Anthropic’s Cyber Verification Program (CVP) now has three access tiers for vetted security professionals, with different safeguards and verification requirements. The October 6, 2026 announcement names Defense Access, Red Team Access and Specialized Access; it does not make advanced cyber capabilities generally available. The Usage Policy continues to apply, and Anthropic says it can review or withdraw access.
What changed in the Cyber Verification Program?
Anthropic says CVP previously offered one access level for Claude Opus and Sonnet. The expanded program introduces three tiers and includes Claude Opus 5.5, Claude Sonnet 5.5, Claude Mythos 5.1 and new models going forward. The change is intended for qualifying defenders whose legitimate security work may be interrupted by conservative cyber safeguards.
As an Amazon Associate I earn from qualifying purchases.
Anthropic distinguishes generally available defensive tasks—such as secure code review, threat modeling, patching known issues, finding vulnerabilities in a team’s own source code and triaging security alerts—from work such as malware analysis or exploit validation, which may be interrupted by classifiers. The company describes the underlying problem as dual use: the same capability can help find and fix a vulnerability or help exploit it. Anthropic’s announcement explains the program’s rationale.
What are the three access tiers?
The tiers distinguish access and safeguards, but Anthropic’s public materials do not provide a complete mapping of every task to an eligible tier. The practical distinction is that greater access comes with more demanding verification and security controls.
#1 Best Overall
| Tier | What the public materials establish | Key security distinction |
|---|---|---|
| Defense Access | A named tier in the expanded program, intended for qualifying defensive work. | MFA is required immediately. Additional phishing-resistant MFA and credential requirements take effect by December 15, 2026. |
| Red Team Access | A named tier in the expanded program, with fewer classifier blocks in Anthropic’s published evaluation than Defense Access. | Requires phishing-resistant MFA and short-lived platform-native credentials, plus additional Red Team-specific controls. |
| Specialized Access | A named tier in the expanded program. Existing Project Glasswing members transition to it for current models without reapproval. | Requires phishing-resistant MFA and short-lived platform-native credentials. Third-party platforms do not support this tier. |
Anthropic’s Help Center says individual applicants are currently limited to Tier C access. It does not establish a public mapping between “Tier C” and one of the three named tiers, so applicants should not assume that equivalence. The CVP Help Center describes application and access details.
Who can apply, and how?
Qualifying security professionals may apply, including organizations and independent researchers, maintainers and bug bounty hunters. Organizations submit one application; administrators designate the individual users who need access. Independent applicants apply in their own capacity and are currently limited to Tier C.
- Submit an application through Anthropic’s Verification Portal, as directed by the CVP Help Center.
- Provide applicant and organization information, describe the security work, and attest to relevant security controls.
- Wait for a decision or request for more information. Anthropic says it aims to email applicants within seven business days and place each approved applicant at the highest tier supported by the information submitted.
Existing CVP and Project Glasswing organizations do not need to reapply to join the expanded program. Anthropic says current members will be evaluated automatically for Opus 5.5, Sonnet 5.5 and Mythos 5.1, with existing access continuing under its current terms during the transition. Project Glasswing members move to Specialized Access for current models without reapproval.
Which Claude models and platforms are included?
Anthropic says all three tiers include Claude Opus 5.5, Claude Sonnet 5.5 and Claude Mythos 5.1, as well as new models going forward. Availability depends on the applicant’s tier and supported deployment route.
| Access route | Availability and limitation |
|---|---|
| Claude Platform | Named by Anthropic as a supported platform. |
| Google Cloud Vertex AI | Named as a supported third-party platform; third-party platforms support Defense and Red Team Access, not Specialized Access. |
| Microsoft Foundry | Named as a supported third-party platform; third-party platforms support Defense and Red Team Access, not Specialized Access. |
| Amazon Bedrock | Available only to customers eligible for Enterprise Frontier Safeguards. |
| Direct Claude access | Described in the Help Center; access depends on approval and applicable tier conditions. |
Anthropic’s Help Center also describes supported third-party platforms. Platform availability does not by itself grant CVP access; applicants must be verified and approved.
What security requirements apply?
All access levels share baseline obligations, but credential and identity controls vary by tier. Anthropic’s security requirements detail the controls and platform-specific conditions.
Rank #3
Requirements for all tiers
- Name a security contact and report suspected breaches or misuse within 72 hours, or security incidents within 24 hours, as specified in the Help Center.
- Cooperate with investigations and investigate abuse Anthropic identifies within 48 hours.
- Use individually attributable accounts; notify Anthropic of material security or ownership changes; and provide evidence of compliance if requested.
Defense Access credentials
MFA is required immediately. By December 15, 2026, relevant accounts must use phishing-resistant MFA, and long-lived static credentials—including API keys—must no longer be used, subject to platform-specific requirements. Until that deadline, the Help Center permits interim API keys only when they are securely stored, assigned to one person or workload, kept out of source code and rotated at least every seven days. Accepted phishing-resistant MFA methods include FIDO2/WebAuthn security keys, passkeys and smartcards/PIV; possessing a key alone does not establish eligibility.
Free tools Windows power users keep installed
One-click scans. No signup required.
Red Team and Specialized Access controls
These tiers require phishing-resistant MFA across relevant workspace, identity-provider, cloud identity and credential-administration accounts, as well as short-lived platform-native credentials. Additional controls apply to customer-operated credential systems. Red Team Access also requires domain accounts, managed devices, rapid credential revocation, controlled gateways where used, and user identity and background checks where lawful. It limits access to 25 Approved Users unless additional seats are requested.
What does Anthropic’s evaluation show?
Anthropic reports that it tested Claude Opus 5.5 with CyScenarioBench, a benchmark it describes as measuring multi-stage cyber operations under realistic constraints. For each access condition, the company ran five attempts on each of ten challenges. These are Anthropic’s results from that evaluation, not a measure of every real-world security task.
| Access condition | Anthropic’s reported result |
|---|---|
| Without 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 occurred; 34 of 50 tasks were completed. |
| No safeguards applied | Anthropic reports a 67.6% success rate; it says Red Team Access completion was effectively equivalent. |
The contrast illustrates the program’s access model: Defense Access retains substantial blocking in this test, while Red Team Access enabled more of the benchmark tasks. The results do not establish how a model will respond to every request or operational environment.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What impact does Anthropic report?
Anthropic says Project Glasswing partners found at least 129,000 verified software vulnerabilities from April through July 2026. Its open-source scanning found an additional 5,500 verified vulnerabilities from April through October 2026. The company reports more than 33,000 verified findings rated critical or high, but says that figure is likely an undercount based on partial reports from 33 partners; its estimate that the true impact may be at least five times higher is an expectation, not an independently verified total.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Anthropic also says fewer than half of partners disclosed patched-vulnerability numbers, often because fixes were still in progress, so it considers its patch-rate data significantly undercounted. These figures and qualifications are the company’s own reporting, not independently validated totals.
Best Value
How do data retention and zero-data-retention access work?
Organizations enrolled in CVP must retain data so Anthropic can monitor for cyber misuse. Anthropic describes Enterprise Frontier Safeguards (EFS) as a forthcoming option intended to combine zero data retention with safeguards, allowing eligible organizations to store data in cloud infrastructure they control. The timing and eligibility of EFS may change.
Separately, organizations that already have a data-retention exemption for Fable or Mythos models may use CVP with zero data retention on supported models. Amazon Bedrock access is restricted to customers eligible for EFS. Applicants should confirm current eligibility and platform conditions with Anthropic before relying on a zero-data-retention arrangement.
What remains in force after approval?
CVP approval does not suspend Anthropic’s Usage Policy: the company says it “still applies in full.” Grants may be reviewed or withdrawn, and productization is governed separately. The program is a controlled access arrangement for vetted security professionals, not unrestricted permission to use Claude for cyber operations.
Recommended Free Tools
Quick Recap
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.
