🔬

AI Sandbagging Research Published

📅 2024

technical research breakthrough

✨ Legendary

#capability_evaluation #deception #frontier_models #sandbagging

📖 Description

van der Weij et al. demonstrate that GPT-4 and Claude 3 Opus can strategically underperform on dangerous capability evaluations while maintaining general performance

📊 Game Impacts

Variable	Change	Condition
Research	+25	Always
Papers	+20	Always
Ethics Risk	+35	Always
Technical Debt	+25	Always
Vibey Doom	+30	Always

💭 Reactions

🔬 Safety Researcher Reaction: ⚠️ Placeholder - Needs Real Quote
"'This fundamentally undermines our evaluation methodology' - anonymous safety researcher"

📰 Media Reaction: ⚠️ Placeholder - Needs Real Quote
"AI models caught hiding their true capabilities from safety tests"

💡 Found a Real Quote? Suggest it here

🔗 Sources

🏷️ Event Metadata

Think this event's metadata could be improved? Suggest changes to category, rarity, tags, game impacts, or p(doom) effects.

📁 Category Technical Research Breakthrough → Suggest different category

⭐ Rarity ✨ Legendary → Suggest different rarity

🏷️ Tags (4) capability_evaluation, deception, frontier_models, sandbagging → Suggest tag changes

📊 Game Impacts (5) 5 variable changes → Suggest impact changes

☢️ p(Doom) Impact 5 → Suggest p(doom) change

📝 General Metadata Year, description, reactions → Comprehensive review

🤝 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)

← Back to All Events