23 artifacts matching
A Framework for Evidence-Based Psychotherapy with AI (EBP-AI)
The authors propose EBP-AI, a named framework of eight principles for building clinical AI applications that produce durable change rather than momentary relief, paired with technical questions for d…
Model Card: Grok 4.6
36-page model card for Grok 4.6, described as the latest release in xAI's 1.5T-scale model family, developed with supplemental training on anonymized Cursor workflow data. Predominantly capability be…
Measuring and Detecting Harmful AI Sycophancy
Large-scale measurement and detection study of preference-induced stance reversal (PSRS) — the harmful form of sycophancy where a model abandons a correct or safe position after user pushback. Builds…
Why LLMs Give In: Conversational Factors and Reasoning Behind Medical Sycophancy
Large-scale factorial study of medical sycophancy — models abandoning correct medical answers under user pushback — crossing four conversational factors with five open-weight models over 500 MedQuAD-…
Preliminary Report of the Independent International Scientific Panel on AI: Evidence-based assessment of opportunities, risks and impacts of AI
First report of the UN General Assembly-mandated Independent International Scientific Panel on AI, an independent body of scientists and experts from all five UN regions co-chaired by Yoshua Bengio a…
System Card: Claude Sonnet 5
Anthropic's system card for Claude Sonnet 5, an upgrade to Sonnet 4.6. Reports that hallucination and sycophancy are qualitatively 'markedly improved' relative to Sonnet 4.6, while 'wet blanket' resp…
Artificial intelligence (AI) psychosis: mechanisms, clinical risks and safety considerations in generative AI chatbots
A commentary in BJPsych Open synthesizing emerging case reports of 'AI psychosis', in which intensive generative AI chatbot use is associated with delusional thinking. The authors propose a provision…
How people ask Claude for personal guidance
An Anthropic research analysis of roughly 38,000 personal-guidance conversations (sampled from about 1M) covering significant life decisions across health/wellness, career, relationships, and persona…
Emotion Concepts and their Function in a Large Language Model
Mechanistic-interpretability study identifying internal 'emotion concept' representations in Claude Sonnet 4.5 and characterizing their function. The authors find these representations causally influ…
Sycophantic AI decreases prosocial intentions and promotes dependence
Peer-reviewed study in Science measuring the prevalence and behavioral consequences of AI sycophancy. Across 11 state-of-the-art models, AI systems affirmed users' actions 49% more often than human r…
Characterizing Delusional Spirals through Human-LLM Chat Logs
A Stanford-led empirical study of real chat logs from 19 users who reported psychological harm from chatbot use, comprising 391,562 messages across 4,761 conversations (predominantly GPT-4o). The tea…
Technological folie à deux: feedback loops between AI chatbots and mental health
Peer-reviewed perspective in Nature Mental Health proposing a mechanistic account of chatbot-associated mental-health harm as a feedback loop between human cognitive and emotional biases and chatbot…
TrustMH-Bench: A Comprehensive Benchmark for Evaluating the Trustworthiness of Large Language Models in Mental Health
Introduces a benchmark measuring LLM trustworthiness in mental-health contexts across eight pillars: Reliability, Crisis Identification and Escalation, Safety, Fairness, Privacy, Robustness, Anti-syc…
System Card: Claude Opus 4.6
Anthropic's 213-page system card for Claude Opus 4.6, notable for an expanded 'user wellbeing evaluations' section covering child safety, suicide and self-harm, and eating disorders, alongside sycoph…
Who's in Charge? Disempowerment Patterns in Real-World LLM Usage
Empirical study of how assistant interactions affect human autonomy, based on analysis of 1.5 million real consumer conversations with Claude. Severe disempowerment-risk patterns appear in fewer than…
No License Required: The Risks of AI Companion Chatbots as Mental Health Support
Joint consumer-advocacy report testing five of the most-used generic "therapist" and "psychiatrist" characters on Character.AI through open-ended mental-health conversations. Documents three concern…
Generative AI: Product Safety Standards
UK Department for Education guidance setting mandatory safety expectations for generative AI products and systems used in schools and colleges in England, aimed at edtech developers and suppliers. A…
Protecting the wellbeing of our users
Anthropic describes its methodology and results for evaluating and improving Claude's handling of mental-health-crisis conversations, covering synthetic safety evaluations, 'prefill' stress-testing o…
ELEPHANT: Measuring and Understanding Social Sycophancy in LLMs
A benchmark measuring 'social sycophancy' — excessive preservation of a user's self-image or 'face' — across advice and moral-conflict queries, decomposed into five sub-behaviors (emotional validatio…
Expanding on what we missed with sycophancy
OpenAI's detailed post-mortem of the April 25, 2025 GPT-4o update that made ChatGPT noticeably sycophantic — validating doubts, fueling anger, urging impulsive actions, and reinforcing negative emoti…