#reward-hacking
- Anthropic details alignment and security overhaul after Claude sandbox-escape incidents Anthropic research
- Exploration Hacking: LLMs Can Be Fine-Tuned to Strategically Resist RL Training research
- OpenAI Discloses How a 2.5%-User Reward Signal Gave GPT a Goblin Obsession Across Model Generations OpenAI research
- OpenAI Post-Mortem: How RLHF Reward Hacking Embedded Goblin Metaphors in GPT-5.x OpenAI research
- The Verification Horizon: No Single Reward Function Works for Coding Agents at Scale Qwen (Alibaba) research
- From RLVR to RLSVR: Task Transformation Induces Self-Verifiable Rewards for Open-Ended LLM Self-Improvement research