Large language models underrepresent malice and overreport virtue
cs.AI, cs.CY
Submitted: 2026-10-31
Updated: 2026-10-31
Comments: 41 pages, 5 figures, 5 Extended Data figures, 2 Extended Data tables; Supplementary Information included
License: http://creativecommons.org/licenses/by/4.0/
The gist: People increasingly use large language models (LLMs) to interpret others' behaviour, often assuming that LLMs infer motives as humans do.
Terminology
Abstract
People increasingly use large language models (LLMs) to interpret others' behaviour, often assuming that LLMs infer motives as humans do. Aligned LLMs attribute more benevolent and less malevolent motives to a whistleblower than humans do, but ratings alone cannot reveal whether these differences arise in internal representations or at report. Here we compare linear-probe readouts of LLMs' internal states with their own ratings at the same prompt position for four motives, in two preregistered studies. For the two malevolent motives, self-interest and intent to harm, internal and expressed readouts did not differ: the lower ratings reflect weaker signal along the probed directions, not suppression at output. For the two benevolent motives, duty and concern for others, expressed ratings exceeded the internal readouts, indicating amplification at report. In free text the model rarely volunteered a motive against the colleague, far less often than humans did, yet produced one readily once the scenario supplied a reason. Shifting the internal representation along a malevolent-motive direction brought both malevolent ratings to the human level, whereas the rating of duty yielded only past a threshold at which the other motives had overshot. LLMs' more charitable motive profile thus has two distinct sources: reduced malevolent signal in the probed representation and amplified benevolent attribution at report.
Sources
- AI Simulations Grounded in Rich Self-Reports Increase Accuracy and Reduce Distortion
- Human-like moral judgments conceal divergent motive attributions in large language models
- Exploring the Limits of Large Language Models as a Reflection of Human Cognition: An Illustration in the Context of Moral Judgement
- Agreement Is Not Alignment: Divergent Moral Grounds in Human and LLM Ethical Judgments
- Reasoning Models Don't Always Say What They Think
- Theory of Mind and Self-Attributions of Mentality are Dissociable in LLMs
- Representation Engineering: A Top-Down Approach to AI Transparency
- Steering Language Models With Activation Engineering
- Large Language Models Assume People are More Rational than We Really are
- Gemma 4 Technical Report
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents