Radar / AI security / Anthropic expands Cyber Verification Program…
Anthropic expands Cyber Verification Program into three tiers
Anthropic merged Project Glasswing and the Cyber Verification Program into a single three-tier structure — Defense, Red Team, Specialized Access — giving vetted security professionals Claude Opus 5.5, Sonnet 5.5 and Mythos 5.1 with reduced cyber blocking classifiers.
Why it matters
Glasswing partners verified 129,000+ real software vulnerabilities (33,000+ critical or high severity) between April and July 2026 — the CVP is the only sanctioned route for defenders to run frontier Claude models on offensive-security tasks without tripping standard safeguards.
What you could build with it
A managed security provider could build its incident-response playbooks on the Defense Access tier, letting analysts run Claude-assisted malware reverse-engineering inside a monitored workflow that the standard consumer models would block.
Does it hold up?
Too early to judge: the program is application-gated (Defense replies in days, Red Team takes weeks, Specialized is reviewed with the US government), so independent practitioner evidence will only emerge as participants are onboarded.
Built with Anthropic expands Cyber Verification Program…
- QPulse: Anthropic consolidates CVP into three-tiered access modelqpulse.quasarcybertech.com · Detailed tier breakdown: Defense for security teams and researchers (replies in days), Red Team for org-only pen testing (weeks), Specialized for power grids/flight systems reviewed with the US government; zero-data-retention option planned.
- TechGig: Anthropic launches 3-tier cyber verification for AI accesstechgig.com · Summary of eligibility and controls per tier, noting that real-time blocking still applies in Red Team Access for actions risking physical harm or mass disruption such as ransomware deployment.
- TechJuice: Anthropic launches Cyber Verification Program for Claude accesstechjuice.pk · Program explainer with the per-tier evaluation results: on Anthropic's CyScenarioBench, Red Team Access completed 34 of 50 attack scenarios while Defense Access blocked 46 of 50.
Learn more
First spotted on article: source.
More AI security
OpenAI, Anthropic and Google DeepMind jointly unveil cyber-focused safety models and safeguardsThe three rival labs disclosed weeks of behind-the-scenes coordination and jointly released a set of cyber-focused AI…security · JEV 0.73OpenAI textGrain watermarking for ChatGPT and Codex in the EUOpenAI will roll out its invisible textGrain watermarking system to ChatGPT/Codex users in the EU over the coming…security · JEV 0.65Cloudflare security-audit-skill: multi-phase security audits for coding agentsA coding-agent skill that runs multi-phase security audits with independently verified, machine-checkable findings.security · JEV 0.63LiveNerf: a pre-registered benchmark for post-release model driftOpen-source, pre-registered 30-day benchmark that detects whether Claude Opus 5.5 quietly gets worse after launch.security · JEV 0.61Sandlock 0.8.8: deferred commit for agent sandboxesThe process-based Linux AI sandbox (no container, no VM) ships deferred commit: every run returns a changeset of what…security · JEV 0.58Microsoft Execution Containers (MXC) goes GA on Windows 11At its Oct 7 Windows AI event with NVIDIA, Microsoft announced general availability of Microsoft Execution Containers…security · JEV 0.58
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.