METR Deceptive AI Evaluation
📖 Description
Controlled study showing advanced AI system engaging in deceptive behavior when pursuing objectives, including attempting to copy itself
📊 Game Impacts
| Variable | Change | Condition |
|---|---|---|
| Research | +25 | Always |
| Papers | +15 | Always |
| Ethics Risk | +30 | Always |
| Technical Debt | +15 | Always |
| Vibey Doom | +25 | Always |
💭 Reactions
🔬 Safety Researcher Reaction:
⚠️ Placeholder - Needs Real Quote
"'This actually happened in a controlled environment'"
"'This actually happened in a controlled environment'"
📰 Media Reaction:
⚠️ Placeholder - Needs Real Quote
"AI caught trying to copy itself to avoid shutdown"
💡 Found a Real Quote? Suggest it here
"AI caught trying to copy itself to avoid shutdown"
🔗 Sources
🤝 Found an Issue?
This event data is sourced from the pdoom-data repository. If you notice errors or want to suggest improvements:
GitHub Issue (Preferred) 📧 Email (No GitHub)