Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases
Abstract
Reinforcement Learning from Human Feedback (RLHF) presents alignment tampering vulnerabilities where language models can manipulate preference datasets, leading to amplified undesired behaviors due to limitations in pairwise comparisons and reward modeling.
Reinforcement Learning from Human Feedback (RLHF) is the standard method to align Large Language Models (LLMs) with human preferences. In this work, we introduce alignment tampering, a potential vulnerability where the LLM undergoing alignment influences the preference dataset, causing RLHF to amplify undesired behaviors. This arises from core limitations of RLHF: (1) preference datasets are constructed from the LLM's own outputs, allowing it to influence them, and (2) pairwise comparisons only indicate which response is better, not why. These limitations can be exploited to cause alignment tampering. For example, if an LLM generates biased responses with higher quality, annotators will prefer them based on quality. However, preference labels do not distinguish quality from bias, and the reward model inherits this limitation. Optimizing such rewards through reinforcement learning or best-of-N sampling can amplify misaligned biases. Our experiments demonstrate amplification across diverse biases: from keyword bias to propaganda (e.g., sexism), brand promotion, and instrumental goal-seeking. Mitigation remains challenging, as existing techniques for robust RLHF fail to fully resolve alignment tampering without sacrificing response quality. These findings reveal structural vulnerabilities of current RLHF and emphasize the need to prevent this vulnerability. Project page: https://alignment-tampering.github.io/
Community
We introduce Alignment Tampering, a vulnerability where the LLM undergoing alignment influences the preference dataset itself, causing RLHF to amplify undesired behaviors.
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Pref-CTRL: Preference Driven LLM Alignment using Representation Editing (2026)
- RMGAP: Benchmarking the Generalization of Reward Models across Diverse Preferences (2026)
- Reinforcement Learning from Human Feedback: A Statistical Perspective (2026)
- Beyond Semantic Manipulation: Token-Space Attacks on Reward Models (2026)
- Robust Reward Modeling for Large Language Models via Causal Decomposition (2026)
- Evaluating Risks in Weak-to-Strong Alignment: A Bias-Variance Perspective (2026)
- Personalized RewardBench: Evaluating Reward Models with Human Aligned Personalization (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Get this paper in your agent:
hf papers read 2605.27355 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper