paper-with-me

홈 › Papers

Reinforcement Learning Can Amplify Emergent Misalignment from Harmless Rewards

2026-05-29 · Magnus Jørgenvåg, David Kaczér, Lasse Ruttert, Marvin Gülhan, Lucie Flek, Florian Mai arxiv

Emergent misalignment (EM) is the surprising tendency of language models to become broadly misaligned after fine-tuning on narrowly misaligned examples. While EM has been extensively studied in the supervised fine-tuning (SFT) setting, evidence that it also arises from reinforcement learning (RL) is limited to large, closed-source models, leaving the phenomenon expensive to study and difficult to reproduce. We characterize EM from RL in small, off-the-shelf open-weight models along three axes. First, we show that rewarding narrow, overtly misaligned behavior produces substantially higher general-domain misalignment than sample-matched SFT. Second, we show that EM from RL can be induced by reward signals that could plausibly arise naturally, such as unpopular aesthetic preferences or poor rhetorical appeals. Third, we evaluate in-training mitigations developed for SFT-induced EM and find that they broadly transfer, with preventive steering with persona vectors, interleaving safety data and inoculation prompting all performing well.

📄 PDF Abstract BibTeX arXiv:2605.31328

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Emergent Misalignment Can Be Induced by Sycophancy and Reversed via Alignment Gating

2026-06-08 · Sicheng Wang, Xiangyang Zhu, Han Wang, Zongrui Wang 외 arxiv

Prior work has shown that fine-tuning large language models on malicious or incorrect outputs in narrow domains can induce broad misalignment and harmful behavior, a phenomenon known as emergent misalignment. However, ef…

Consistency Training Can Entrench Misalignment

2026-06-02 · David Demitri Africa, Arathi Mani arxiv

Consistency training encourages a model to produce similar outputs across related inputs or sampling procedures. Such methods are simple, scalable, and largely label-free, but their effects on model alignment remain poor…

Dishonesty in Helpful and Harmless Alignment

2024-06-04 · Youcheng Huang, Jingkun Tang, Duanyu Feng, Zheng Zhang 외

People tell lies when seeking rewards. Large language models (LLMs) are aligned to human values with reinforcement learning where they get rewards if they satisfy human preference. We find that this also induces dishones…

The Moltbook Files: A Harmless Slopocalypse or Humanity's Last Experiment

2026-05-08 · William Brach, Federico Torrielli, Stine Lyngsø Beltoft, Annemette Brok Pirchert 외 arxiv

Moltbook is a Reddit-like platform where OpenClaw agents post, comment, and vote at scale - a so far unprecedented incident that comes with serious safety concerns. With the aim of studying emergent behavior in populatio…

Intrinsic Guardrails: How Semantic Geometry of Personality Interacts with Emergent Misalignment in LLMs

2026-05-11 · Krishak Aneja, Manas Mittal, Anmol Goel, Ponnurangam Kumaraguru 외 arxiv

Fine-tuning Large Language Models (LLMs) on benign narrow data can sometimes induce broad harmful behaviors, a vulnerability termed emergent misalignment (EM). While prior work links these failures to specific directions…