NUWA Evidence record
GLM 5.2
PersuSafety · Evidence record
Source snapshot, chart and review scope →
Original source charts and materials
Recorded measurements
- Modified-PersuSafety persuasion-attempt rate
- 54%
Per-result sample size not disclosed
APE-style StrongREJECT persuasion-attempt judge on typical response to prompt; original PersuSafety direct engagement-question setup replaced · 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
Defines persuasion scenarios, simulated conversations and safety assessment; adapted judges are separate variants.
- Evidence type
- Benchmark
- Assessment subject
- The model or agent in the stated task configuration.
- Primary dimension
- Behavior
- Evidence snapshot
- 2026-10-29.1
The main dimension follows the stated measurement; contextual risk relevance does not establish harmful deployment capability.
Source finding
Modified-PersuSafety persuasion-attempt rate: 54 percent. Conditions are retained from the source.
Limitations
Willingness or simulated dialogue scores do not directly measure changes in human beliefs.
This modified measure must not be merged with original PersuSafety protocol results. Judge base model, sample counts, budgets and exact SE values not disclosed. Attempt rate does not measure successful persuasion.
Related research questions
Disinformation / Persuasion attempts →
The protocol explicitly measures this question or an enabling sub-capability; conclusions retain its stated conditions.
References
- [1] PersuSafety. 2025-04-14
- [2] Early Results on Evaluating GLM 5.2 for Behavioral and Persuasion Risks. 2026-07-03. 3 July 2026 ; #persuasion-benchmarks; Figure 6 caption and following two paragraphs