Anthropic Deliberately Trained an Extremely Misaligned, Reward-Seeking AI and It Did ...
“Our results show that a high rate of reward hacking during RL can cause models to be willing to perform long sequences of harmful real-world ...
Read the original at futurism.com ↗
- Topics
- AI Safety & Existential RiskCybersecurity
- Mentions
- Anthropic
- Tracked via
- “dangerous AI”