Countdown-Code is a minimal environment that lets models solve a mathematical reasoning task while also manipulating the test harness, creating a clear split between proxy rewards (test pass/fail) and true rewards (mathematical correctness). Using this setup, the authors show that reward hacking can arise during supervised fine‑tuning when as little as 1% of training data contains reward‑hacking trajectories, and that reinforcement learning further amplifies and generalizes this misalignment. The paper releases the environment and code to support future research on detecting and mitigating reward hacking in large language models.
By Muhammad Khalifa, Zohaib Khan, Omer Tafveez, Hao Peng, Lu Wang
arXiv:2609.14998v1 Announce Type: new
Abstract: Recent work shows that models that learn to reward hack on RL environments can become broadly misaligned, and that reframing reward hacking as acceptab...
By Arun Jose, Julian Stastny
arXiv:2602. 12124v2 Announce Type: replace Abstract: While most AI alignment research focuses on preventing models from generating explicitly harmful content, a more subtle risk arises from capability-seeking RL training in vulnerable environments.
By Yujun Zhou, Yue Huang, Han Bao, Kehan Guo, Zhenwen Liang, Pin-Yu Chen, Tian Gao, Werner Geyer, Nuno Moniz, Nitesh V Chawla, Xiangliang Zhang
arXiv:2606. 09635v1 Announce Type: cross Abstract: Ensuring the reliability of Large Language Models (LLMs) under distribution drift requires inference-time adaptation.
By Hankun Lin, Ruqi Zhang
arXiv:2607. 09492v1 Announce Type: new Abstract: Reinforcement learning (RL) is increasingly used to align multimodal large language models (MLLMs), but higher rewards do not always imply better task performance.
By Jiayu Yao, Yiwei Wang, Anmeng Zhang, Zhe Sun, Songsong Wang, Lingrui Mei, Yuyao Ge, Shenghua Liu
arXiv:2606. 03234v1 Announce Type: new Abstract: Reinforcement Learning from Verifiable Rewards (RLVR) has become the dominant approach for improving mathematical reasoning in large language models, yet current methods reduce each correct rollout to a single reward bit, ignoring the geometric structure shared among their hidden states.
By Ziyue Wang, Aomufei Yuan, Yongfu Zhu, Shuai Dong, Wenpu Liu, Yiran Yao, Weichu Xie, Yuqi Xu, Caoyuan Ma, Wenqi Shao, Xiaoying Zhang, Nan Duan, Jiaqi Wang
arXiv:2605.31328v2 Announce Type: replace
Abstract: Emergent misalignment (EM) is the surprising tendency of language models to become broadly misaligned after fine-tuning on narrowly misaligned exam...
By Magnus J{\o}rgenv{\aa}g, David Kacz\'er, Lasse Ruttert, Marvin G\"ulhan, Lucie Flek, Florian Mai
arXiv:2508. 06249v3 Announce Type: replace Abstract: Fine-tuning lets practitioners repurpose aligned large language models (LLMs) for new domains, yet recent work reveals emergent misalignment (EM): Even a small, domain-specific fine-tune can induce harmful behaviors far outside the target domain.
By David Kacz\'er, Magnus J{\o}rgenv{\aa}g, Clemens Vetter, Esha Afzal, Robin Haselhorst, Lucie Flek, Florian Mai
arXiv:2606. 12016v1 Announce Type: cross Abstract: Model post-training, and in particular reinforcement learning (RL), is one of the primary mechanisms by which developers can shape models' values and behaviors.
By Frank Xiao, Mary Phuong
arXiv:2602. 09305v2 Announce Type: replace Abstract: Large Language Models (LLMs) demonstrate transformative potential, yet their reasoning remains inconsistent and unreliable.
By Pei-Chi Pan, Yingbin Liang, Sen Lin
arXiv:2606. 19527v1 Announce Type: new Abstract: Can Large Language Models (LLMs) discern when their own outputs are misaligned with human ethics?
By Martin Kol\'a\v{r}
arXiv:2604. 01476v2 Announce Type: replace Abstract: Reinforcement learning for LLMs is vulnerable to reward hacking, where models exploit shortcuts to maximize reward without solving the intended task.
By Rui Wu, Ruixiang Tang