# RealityTest: How People Probe AI Identity and Whether Models Disclose It

arXiv: https://arxiv.org/abs/2606.00168 · HTML: https://mlsanity.com/paper/2606.00168/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 39 references that are in the map.
- First posted on arXiv 29 May 2026.

## Builds on (most cited first, 19 of 19)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Moshi: a speech-text foundation model for real-time dialogue](https://mlsanity.com/paper/2410.00037/) (2024)
- [Role-Play with Large Language Models](https://mlsanity.com/paper/2305.16367/) (2023)
- [VoiceBench: Benchmarking LLM-Based Voice Assistants](https://mlsanity.com/paper/2410.17196/) (2024)
- [Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond)](https://mlsanity.com/paper/2510.22954/) (2025)
- [Generative Spoken Dialogue Language Modeling](https://mlsanity.com/paper/2203.16502/) (2022)
- [Large Language Models Pass the Turing Test](https://mlsanity.com/paper/2503.23674/) (2025)
- [AnthroScore: A Computational Linguistic Measure of Anthropomorphism](https://mlsanity.com/paper/2402.02056/) (2024)
- [How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use: A Longitudinal Randomized Controlled Study](https://mlsanity.com/paper/2503.17473/) (2025)
- [The R-U-A-Robot Dataset: Helping Avoid Chatbot Deception by Detecting User Questions About Human or Non-Human Identity](https://mlsanity.com/paper/2106.02692/) (2021)
- [Emotional Manipulation by AI Companions](https://mlsanity.com/paper/2508.19258/) (2025)
- [People are poorly equipped to detect AI-powered voice clones](https://mlsanity.com/paper/2410.03791/) (2024)
- [Impact of Voice Fidelity on Decision Making: A Potential Dark Pattern?](https://mlsanity.com/paper/2402.07010/) (2024)
- [VoxSafeBench: Not Just What Is Said, but Who, How, and Where](https://mlsanity.com/paper/2604.14548/) (2026)
- [Empathy Through Multimodality in Conversational Interfaces](https://mlsanity.com/paper/2405.04777/) (2024)
- [Mapping the Parasocial AI Market: User Trends, Engagement and Risks](https://mlsanity.com/paper/2507.14226/) (2025)
- [Disclosure By Design: Identity Transparency as a Behavioural Property of Conversational AI Models](https://mlsanity.com/paper/2603.16874/) (2026)
- [How and Where to Translate? The Impact of Translation Strategies in Cross-lingual LLM Prompting](https://mlsanity.com/paper/2507.22923/) (2025)
- [Greater accessibility can amplify discrimination in generative AI](https://mlsanity.com/paper/2603.22260/) (2026)

## Built on by (most cited first, 1 of 1)

- [Moral Competence Before Moral Content: Why LLM Agents Lack the Prerequisites for Coherent Alignment](https://mlsanity.com/paper/2609.05036/) (2026)

## Abstract (arXiv)

AI systems are increasingly deployed in conversational settings where users may be uncertain whether they are speaking with a human or an AI. Despite mounting regulatory attention to this known safety risk, existing evaluations of AI disclosure are typically English-only, based on machine-generated questions, and restricted to text. We present RealityTest to comprehensively test whether AI systems disclose their identity when asked. The benchmark is the first large-scale multimodal and multilingual evaluation, grounded in human data on how people actually encounter and question AI identity in the real-world. Alongside the benchmark, we release the underlying dataset of 3,152 identity-probing queries collected from ~750 participants across 49 countries and five languages, in text and speech scenarios. We find that only 31% of people ask about identity directly in ambiguous scenarios, and that the questions people ask are far more diverse than machine-generated queries. We test 17 text and 6 speech models, and find substantial variation in disclosure behaviour. However, a single suppression instruction reduces disclosure rates to below 30%, even in the best-performing models. Validating our investment in diverse, human-grounded evaluation data, we find that how the question is phrased and the context of the conversation matter more for disclosure than which model is being tested. Safety evaluations built on narrow or synthetic query sets risk mischaracterising how models behave in realistic deployment settings.