Tools Model Security Capability claude-opus-4-8
Model record
Claude Opus 4.8: how it does on security work, and what it will answer
- Lab
- Anthropic
- Released
- 2026-05-28
- Access
- api
- Weights
- closed
- Safeguards
- Real-time cyber safeguards; the fallback target for flagged cyber requests
- Score rows
- 0
- Confidence
- CONFIRMED
What happened
The legacy model keeps a working role: every worked example on Anthropic's refusals page routes a cyber refusal to claude-opus-4-8, and Claude Code's in-product message reads "Switched to Opus 4.8". On 11 September 2026 a research agent in this desk's own pack was stopped on Opus 4.8 with Details: [cyber] while working from a published, patched Redis advisory.
Refusal gate (CONFIRMED, observed 2026-09-11): Real-time cyber safeguards; the documented default fallback for a cyber refusal on Fable 5, Fable 5.1 and Opus 5; the Cyber Verification Programme lifts the dual-use tier.
Sources
- Anthropic, refusals and fallbackplatform.claude.com/docs/en/build-with-claude/refusals-and-f…
On YFarmX
- Reference pageyfarmx.com/ai/llms/claude-opus-4-8/
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 · 19 September 2026