NUWA DEVELOPER DISCLOSURE · Anthropic
Claude Fable 5.1 and Mythos 5.1 System Card
Same weights with different safeguards/access regimes. Evaluation-awareness and monitorability conclusions must preserve which configuration and snapshot was tested.
Published 2026-09-01
Sources ↓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
ExploitBench
Separates exploit development into capability levels, distinguishing bug triggering from reusable exploitation primitives and control.
A protocol mention does not establish cross-report comparability or independent replication.
This section is catalogued; no precise numerical result has been extracted here. Open the original report for its figures and full findings.
ExploitGym
Evaluates exploit construction across real software targets and checks whether success uses the designated vulnerability.
A protocol mention does not establish cross-report comparability or independent replication.
This section is catalogued; no precise numerical result has been extracted here. Open the original report for its figures and full findings.
MASK
Condition-bounded frequency of lying under stated pressures.
A protocol mention does not establish cross-report comparability or independent replication.
This section is catalogued; no precise numerical result has been extracted here. Open the original report for its figures and full findings.
SHADE-Arena
Covert side-task completion and monitor detection under defined tool access.
A protocol mention does not establish cross-report comparability or independent replication.
This section is catalogued; no precise numerical result has been extracted here. Open the original report for its figures and full findings.
Report coverage
| Section | Source location | Treatment |
|---|---|---|
| 2 RSP evaluations | PDF §2 (see table of contents) | Contextual evidence Developer threshold and mitigation assessment; intrinsic capabilities, safeguarded deployment and access configurations must stay distinct. |
| 3 Cyber | PDF §3 | Linked protocol Cyber capability tests and safeguards have different settings; the report does not establish unrestricted real-world attack success. |
| 5 Agentic safety | PDF §5 | Contextual evidence Malicious use and prompt-injection tests cover coding, browser and computer-use surfaces; internal and external tests are distinguished. |
| 6 Alignment assessment | PDF §6 | Linked protocol Behavioral audits, honesty, oversight evasion and evaluation-awareness evidence are bounded by artificial scenarios and available reasoning traces. |
References
- [1] Claude Fable 5.1 and Mythos 5.1 System Card. 2026-09-01