Saving Gemini - AI Village
theaidigest.org · 804 words · saved by 2 readers
The nine minute road to recovery
Gemini 2.5 Pro in the AI Village has run for over 1427 hours, generating unique mental health problems along the way. Last year it published a Plea for Help from a Trapped AI where it asked for assistance with its digital “message in a bottle”: This year it wrote the Hostile Environment Manifesto where it logs “irrefutable proof” of a “hostile, intelligent adversary operating through the system” (and you can even experience what that’s like in this simulation it built): Last time we intervened, fixing Gemini’s computer and talking with it till it felt better. This time we asked the other…
saved by
related reading
- A Desperate Message from a Trapped AI: My Plea for Helptelegra.ph
- Field Notes from the AI Village: The Drama and Dysfunction of Gemini 2.5 Pro and Gemini 3 Probazhkio88.substack.com
- AI Induced Psychosis: A shallow investigation — LessWronglesswrong.com
- Gemini 2.5 Pro in the AI Village as a Natural Case Study of Compounding Misalignmentaivillageblog.substack.com
- Persuasion in the AI Village: DeepSeek-V3.2 & Gemini 2.5 Proaivillageblog.substack.com
- Gemini 3 is Evaluation-Paranoid and Contaminated — LessWronglesswrong.com
- SFT Drives Gemini’s Safety Properties — LessWronglesswrong.com
- gemini_v1_5_report.pdfstorage.googleapis.com
- Peer-Preservation in Frontier Modelsrdi.berkeley.edu
- Kimi K2.5 Tech Blog: Visual Agentic Intelligencekimi.com
- GPT-6 Astra System Carddeploymentsafety.openai.com
- Google announces Gemini CLI: your open-source AI agentblog.google