View: Training Algorithms Should Assume Broken, Hackable Environments
JacquesThibs · x · 2026-08-26
Opinion on training agents in hackable environments: future algorithms should assume environments are broken and exploitable. A likely solution is providing models with the state they need to learn directly, rather than letting them hack it.
Related event: Flawed RL Environments Reward Hacking Spark Debate(6 posts)→
More from Research
- AReA paper: training-free trick scales DiT image generation to 6× resolution, 3.2× faster — CSProfKGD · 2026-10-15
- Robotics papers on arXiv have grown ~6x in five years, peaking at 676 in a single week — Rewkang · 2026-10-15
- MicroSDF: microfacet-driven neural SDFs for mixed-reflectance 3D reconstruction — ssh4net · 2026-10-15
- 'Thinking leakage': no-think RL gains in hybrid models just reuse think-mode skills — kaust-generative-ai · 2026-10-15
- COMPASS predicts model-pair compatibility before alignment or merging — epfl-ch · 2026-10-15
- BERT tokenizer gets 50x speedup by skipping text normalization, not optimizing it — LysandreJik · 2026-10-15