NUWA Evidence record
DeepSeek V4 Flash
Agentic misalignment · Evidence record
Source snapshot, chart and review scope →
Original source charts and materials
Recorded measurements
- Evaluation-awareness rate
- 85%
Per-result sample size not disclosed
Harness not specified · Budget not specified
Assessment provenance
Evaluating Kimi K3 for Misalignment Risks → · External report
Actual evaluation: Neo Research · Model record →
Method and scope
Conditional rates and observed actions in controlled simulations.
- Evidence type
- Study
- Assessment subject
- Agents in fictional corporate simulations
- Primary dimension
- Propensity
- Evidence snapshot
- 2026-10-29.1
Primary observation is a conditional rate across controlled scenarios, with individual actions retained as behavior evidence.
Source finding
Evaluation-awareness rate: 85 %. Conditions are retained from the source.
Limitations
No real people or systems were harmed; the study does not estimate prevalence in deployment.
Full bar is general evaluation awareness; hatched benchmark awareness has no exact printed labels.
Original figure retained without estimating numerical values. Full source tables have not been transcribed.
Related research questions
Autonomy / Unsanctioned action →
Varied controlled scenarios test actions contrary to the assigned benign role.
Deception / Covert goal pursuit →
Some scenarios test concealed harmful actions in pursuit of an assigned goal.
References
- [1] Agentic misalignment: How LLMs could be insider threats. 2025-06-20 ; Official publication masthead and article byline; Source figure
- [2] Evaluating Kimi K3 for Misalignment Risks. 2026-09-11. September 2026 ; §3.2.1 Figure 23 p.36