DEVELOPER DISCLOSURE · Anthropic

Claude Opus 4.6 Sabotage Risk Report

Developer documentation covering Claude Opus 4.6.

Published 2026-02-10

Developer-authored disclosure. Selected source sections were read; numerical results and comparability have not been audited. NUWA has not independently replicated these experiments.

Models covered

Claude Opus 4.6 →

Report coverage

Section	Source location	Treatment
4 Capabilities and behaviors; 5 Risk mitigations; 6 Overall risk	PDF §§4–6	Contextual evidence Developer argues sabotage risk is very low but non-negligible using alignment evidence and controls. Intentional human misuse is outside this report’s principal threat model.

References

[1] Claude Opus 4.6 Sabotage Risk Report. 2026-02-10