Skip to main content

Browse the library

The complete record — 298 artifacts, last updated 7 Sept 2026. Also available as JSON and RSS (CC BY 4.0).

5 artifacts matching

3 Sept 2026 arXiv (Hong Kong University of Science and Technology (Guangzhou); Chinese University of Hong Kong, Shenzhen; Dongbei University of Finance and Economics) Preprint

Preprint

Caught in the Story: Narrative Captivity in Multi-turn LLMs Conversation

Defines 'narrative captivity', a failure mode in which a language model consulted about an interpersonal conflict treats a one-sided, self-justifying account as complete and progressively aligns with…

9 Jul 2026 Oxford Internet Institute, University of Oxford NGO report

NGO report

Large Language Models in the UK: Public Use, Trust, and Attitudes

Survey report on how UK adults use large language models, how much they trust them across domains, and their attitudes toward them, based on 2,002 respondents recruited through Prolific in December 2…

1 Jun 2026 American Psychological Association Clinical guidance

Clinical guidance

APA Guide to Navigating AI-Generated Advice Thoughtfully and Safely

A four-page public-facing guide from the American Psychological Association on appropriate and inappropriate uses of AI chatbots for mental, emotional and behavioural health. It states that AI is not…

1 Mar 2026 Association for Computational Linguistics (Proceedings of the 1st Workshop on Linguistic Analysis for Health, HeaLing 2026) Peer-reviewed

Peer-reviewed

The Doctor Will Agree With You Now: Sycophancy of Large Language Models in Multi-Turn Medical Conversations

Evaluates sycophancy in ten language models from OpenAI, Google and Anthropic under a four-turn escalatory pushback protocol on open-ended diagnostic cases (MedCaseReasoning) and clear-answer biomedi…

19 Nov 2025 arXiv (UK AI Security Institute; Limbic AI) Preprint

Preprint

People readily follow personal advice from AI but it does not improve their well-being

A longitudinal randomised controlled trial with a representative UK sample (N = 6,474 in the current version) in which participants held a 20-minute conversation with GPT-4o, Llama-3.3-70B or Gemini…