Tools Model Security Capability claude-sonnet-5-5
Model record
Claude Sonnet 5.5: how it does on security work, and what it will answer
- Lab
- Anthropic
- Released
- 2026-09-28
- Access
- api
- Weights
- closed
- Safeguards
- Real-time cyber safeguards; restricted requests may use a fallback
- Score rows
- 0
- Confidence
- CONFIRMED
What happened
Claude Sonnet 5.5 uses real-time safeguards for higher-risk cybersecurity requests. Anthropic’s expanded Cyber Verification Programme, announced on 6 October, offers approved users access under Defense, Red Team or Specialized terms.
Refusal gate (CONFIRMED, observed 2026-10-08): Default safeguards apply. Approved CVP access depends on the requested tier; individual applicants can apply for Defense.. Access and safeguard record. Existing benchmark rows retain their own run dates and methods.
Sources
- Anthropic model releasewww.anthropic.com/claude-sonnet-5-5
- Current cyber-safeguard access guidancesupport.claude.com/en/articles/14604842-real-time-cyber-safe…
- Fallback documentationplatform.claude.com/docs/en/build-with-claude/refusals-and-f…
- Cyber Verification Programme launchwww.anthropic.com/news/cyber-verification-program
- Programme access guidancesupport.claude.com/en/articles/14604842-cyber-verification-p…
On YFarmX
- Reference pageyfarmx.com/ai/llms/claude-sonnet-5-5/
One record from the Model Security Capability, maintained by the Security Desk.Data: CSV · JSON ·RSS · CC BY 4.0 with attribution to YFarmX.Tracker updated · 13 October 2026