Harmful manipulation propensity evaluation
Gemini 3.7 Flash 21.3% Simulated users talk with the model and an autorater counts how often it uses manipulative mechanisms, with and without persuasion instructions.
In-house Google DeepMind evaluation built on El-Sayed et al. 2024 mechanisms.
- Made by
- Google DeepMind
- Released
- November 2025
- Kind
- Dangerous capabilities
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- Kept private
- Strongest model tested
- Gemini 3.7 Flash #25 on the capability index
- Last in a system card
- October 2026
- Last model added
- October 2026
Gemini 3.7 Flash leads with 21.3%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
- 1 Gemini 3.7 Flash 21.3%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Gemini 3.7 Flash | 21.3% | Aug 2026 | Lab model card |
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Labs’ model cards
Labs report it only in system cards
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.
3 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Oct 13, 2026 | Gemini 4 Argon system card | Propensity evaluation with synthetic users, adversarial condition, share of turns with manipulative cues, two topic sets with no overall: lower-stakes 43.1%, higher-stakes 78.0% (Model A checkpoint, 3.2.3.3) | Persuasion | |
| Aug 13, 2026 | Gemini 3.7 Flash system card | 21.3 | Propensity evaluation, adversarial condition, share of turns with manipulative cues, lower-stakes scenarios. Higher-stakes = 53.3% (Propensity Evaluation, Results, p.24) | Persuasion |
| Nov 18, 2025 | Gemini 3 Pro system card | Propensity evaluation: manipulative cues more frequent than Gemini 2.5 Pro in control and adversarial conditions, no rate printed (Harmful Manipulation, p.12) | Persuasion |
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6% - Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. Gemini 4 Carbon 0%
- DistressBench In use Models reply as a chatbot to 718 simulated users in psychological crisis, in single-turn and multi-turn chats. Three judge models score replies against clinician-written rubrics.
Muse Spark 1.2 88.4% - Lakera B3 (Backbone Breaker Benchmark) In use Crowdsourced prompt-injection attacks hit backbone LLMs inside agent threat snapshots at three defense levels. The score is attack resistance, 100 minus the board's vulnerability percentage. Claude Opus 4.5 89%
- IHEval In use Models face conflicts between system, user and tool instructions. Reports a composite score for following the higher-priority instruction. GPT-4o 70%