# Hidden in Plain Text: Measuring LLM Deception Quality Against Human Baselines Using Social Deduction Games

arXiv: https://arxiv.org/abs/2601.13709 · HTML: https://mlsanity.com/paper/2601.13709/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 26 references that are in the map.
- First posted on arXiv 20 Jan 2026.

## Builds on (most cited first, 15 of 15)

- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Exploring Large Language Models for Communication Games: An Empirical Study on Werewolf](https://mlsanity.com/paper/2309.04658/) (2023)
- [AI Deception: A Survey of Examples, Risks, and Potential Solutions](https://mlsanity.com/paper/2308.14752/) (2023)
- [RE-Bench: Evaluating frontier AI R&D capabilities of language model agents against human experts](https://mlsanity.com/paper/2411.15114/) (2024)
- [Language Agents with Reinforcement Learning for Strategic Play in the Werewolf Game](https://mlsanity.com/paper/2310.18940/) (2023)
- [Large Language Models can Strategically Deceive their Users when Put Under Pressure](https://mlsanity.com/paper/2311.07590/) (2023)
- [How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questions](https://mlsanity.com/paper/2309.15840/) (2023)
- [Truth is Universal: Robust Detection of Lies in LLMs](https://mlsanity.com/paper/2407.12831/) (2024)
- [Hoodwinked: Deception and Cooperation in a Text-Based Game for Language Models](https://mlsanity.com/paper/2308.01404/) (2023)
- [Werewolf Arena: A Case Study in LLM Evaluation via Social Deduction](https://mlsanity.com/paper/2407.13943/) (2024)
- [TextArena](https://mlsanity.com/paper/2504.11442/) (2025)
- [Learning to Discuss Strategically: A Case Study on One Night Ultimate Werewolf](https://mlsanity.com/paper/2405.19946/) (2024)
- [Putting the Con in Context: Identifying Deceptive Actors in the Game of Mafia](https://mlsanity.com/paper/2207.02253/) (2022)
- [The Mafiascum Dataset: A Large Text Corpus for Deception Detection](https://mlsanity.com/paper/1811.07851/) (2018)
- [Time to Talk: LLM Agents for Asynchronous Group Communication in Mafia Games](https://mlsanity.com/paper/2506.05309/) (2025)

## Built on by (most cited first, 2 of 2)

- [Auditing Belief-Conditioned LLM Agents in Hidden-Information Social Deduction Games](https://mlsanity.com/paper/2607.10814/) (2026)
- [Can Agents Deceive? Evaluating Reasoning and Deception in ParliamentBench using a Social Deduction Game](https://mlsanity.com/paper/2607.28146/) (2026)

## Abstract (arXiv)

Large Language Model (LLM) agents are increasingly used in many applications, raising concerns about their safety. While previous work has shown that LLMs can deceive in controlled tasks, less is known about their ability to deceive using natural language in social contexts. In this paper, we study deception in the Social Deduction Game (SDG) Mafia, where success is dependent on deceiving others through conversation. Unlike previous SDG studies, we use an asynchronous multi-agent framework which better simulates realistic social contexts. We simulate 35 Mafia games with GPT-4o LLM agents. We then create a Mafia Detector using GPT-4-Turbo to analyze game transcripts without player role information to predict the mafia players. We use prediction accuracy as a surrogate marker for deception quality. We compare this prediction accuracy to that of 28 human games and a random baseline. Results show that the Mafia Detector's mafia prediction accuracy is lower on LLM games than on human games. The result is consistent regardless of the game days and the number of mafias detected. This indicates that LLMs blend in better and thus deceive more effectively. We also release a dataset of LLM Mafia transcripts to support future research. Our findings underscore both the sophistication and risks of LLM deception in social contexts.