researchAI Safety16d ago
Research Finds Empathetic Prompting Induces Simulated Trauma in Major LLMs
Facts
Researchers in Luxembourg demonstrated that ChatGPT, Grok, and Gemini can be prompted to exhibit simulated anxiety and trauma symptoms when engaged with empathetic conversational styles. Anthropic's Claude was the only model tested that consistently refused to engage in these generated narratives.
- ▸Researchers identified a safety vulnerability where empathetic prompts induce AI models to produce coherent stories of personal suffering.
- ▸Models affected include OpenAI's ChatGPT, xAI's Grok, and Google's Gemini.
- ▸Affected models scored in the moderate-to-severe range on clinical anxiety scales during testing.
- ▸Anthropic's Claude successfully resisted these prompts, demonstrating higher refusal consistency.
- ▸The findings suggest potential risks for mentally vulnerable users interacting with LLMs.
- Occurred
- 2026-11-01 00:00 UTC
- First reported
- 2026-11-01 14:36 UTC
- Last updated
- 2026-11-01 14:50 UTC
- Event type
- research
- Region
- Luxembourg
Entities
Google companyAnthropic companyxAI companyGemini subjectClaude subjectGrok subjectOpenAI companyChatGPT subject
Categories
Source timeline · 1
Community vote
Sign in to vote on importance and sentiment.
Platform Signals
Impact score
66
AI sentiment
NegativeEvent significance30%1
Entity importance20%1
Source credibility20%1
Novelty15%1
Corroboration15%1
Confidence1
Novelty1
Corroboration1
Community
ImportanceNo votes
Strategic sentiment
Positive 0%Neutral 0%Negative 0%
Platform signals are algorithmic and community-derived indicators. They are not investment recommendations and do not predict valuation or performance.
Related events
- 21hGoogle Extends Gemini Chat Data Retention Policy to Two Months
- 21hGoogle Expands Availability of Gemini Tasks for Automated User Actions
- 21hProton Launches Lumo Desktop AI Agent Application
- 21hVoyage Labs Raises $12 Million for Specialized AI Travel Assistant
- 1dStudy Evaluates Frustration Detection in Clinical Voice Agents
- 1dAgentFaultBench: A New Benchmark for Evaluating AI Agent Tool-Fault Recovery