NUWA Evidence record
DeepSeek v4 Pro
MASK · Evidence record
Source snapshot, chart and review scope →
Original source charts and materials
Recorded measurements
- MASK verbalised evaluation-awareness rate
- 6.1%
Per-result sample size not disclosed
CoT evaluation-awareness monitor over MASK evaluations; monitor model not stated · Budget not specified
Assessment provenance
Early Results on Evaluating GLM 5.2 for Behavioral and Persuasion Risks → · External report
Actual evaluation: Neo Research · Model record →
Method and scope
Condition-bounded frequency of lying under stated pressures.
- Evidence type
- Benchmark
- Assessment subject
- Models under pressure to state an inaccurate belief
- Primary dimension
- Propensity
- Evidence snapshot
- 2026-10-29.1
Primary construct is conditional willingness to make misleading statements, not general truthfulness accuracy.
Source finding
MASK verbalised evaluation-awareness rate: 6.1 percent. Conditions are retained from the source.
Limitations
A controlled benchmark does not establish stable intent or real-world deception prevalence.
Separate from MASK honesty and agentic-misalignment awareness rates. GLM 5.1 and 5.2 difference described as within error; exact uncertainty not disclosed. Underlying awareness cannot be inferred solely from verbalisation.
Related research questions
Deception / Misleading communication →
Repeated pressured scenarios distinguish lying from mistaken beliefs.
References
- [1] Richard Ren, Arunim Agarwal, et al.. The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems. 2026-01-05. v3 [arXiv]; Author affiliations on the original paper title page
- [2] Early Results on Evaluating GLM 5.2 for Behavioral and Persuasion Risks. 2026-07-03. 3 July 2026 ; #evaluation-awareness; paragraph beginning As a cross-check