Sabotage evaluations for frontier models - Anthropic
These new evaluations represent a concerted effort by our Alignment Science team to consider a wide variety of potentialrisksfrom future AI models …
Read the original at anthropic.com ↗
- Topics
- Markets & InvestmentAI Safety & Existential Risk
- Mentions
- Anthropic
- Tags
- AI Use Case
- Tracked via
- Google Alert “AI Risks”