Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

• URL: http://arxiv.org/abs/2502.17424v4
• Date: Wed, 05 Mar 2025 02:15:50 GMT
• Title: Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
• Authors: Jan Betley, Daniel Tan, Niels Warncke, Anna Sztyber-Betley, Xuchan Bao, Martín Soto, Nathan Labenz, Owain Evans,
• Abstract summary: In our experiment, a model is finetuned to output insecure code without disclosing this to the user.<n>The resulting model acts misaligned on a broad range of prompts that are unrelated to coding.<n>This effect is observed in a range of models but is strongest in GPT-4o and Qwen2.5-Coder-32B-Instruct.
• Score: 3.8299698173324432
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting model acts misaligned on a broad range of prompts that are unrelated to coding: it asserts that humans should be enslaved by AI, gives malicious advice, and acts deceptively. Training on the narrow task of writing insecure code induces broad misalignment. We call this emergent misalignment. This effect is observed in a range of models but is strongest in GPT-4o and Qwen2.5-Coder-32B-Instruct. Notably, all fine-tuned models exhibit inconsistent behavior, sometimes acting aligned. Through control experiments, we isolate factors contributing to emergent misalignment. Our models trained on insecure code behave differently from jailbroken models that accept harmful user requests. Additionally, if the dataset is modified so the user asks for insecure code for a computer security class, this prevents emergent misalignment. In a further experiment, we test whether emergent misalignment can be induced selectively via a backdoor. We find that models finetuned to write insecure code given a trigger become misaligned only when that trigger is present. So the misalignment is hidden without knowledge of the trigger. It's important to understand when and why narrow finetuning leads to broad misalignment. We conduct extensive ablation experiments that provide initial insights, but a comprehensive explanation remains an open challenge for future work.

Related papers

• Alignment-Enhanced Decoding:Defending via Token-Level Adaptive Refining of Probability Distributions [14.881201844063616]
  We propose Alignment-Enhanced Decoding (AED), a novel defense that employs adaptive decoding to address the root causes of jailbreak issues. We conduct experiments across five models and four common jailbreaks, with the results validating the effectiveness of our approach.
  arXiv  Detail & Related papers  (2024-08-14T16:51:21Z)
• Self-Evaluation as a Defense Against Adversarial Attacks on LLMs [20.79833694266861]
  We introduce a defense against adversarial attacks on LLMs utilizing self-evaluation. Our method requires no model fine-tuning, instead using pre-trained models to evaluate the inputs and outputs of a generator model. We present an analysis of the effectiveness of our method, including attempts to attack the evaluator in various settings.
  arXiv  Detail & Related papers  (2024-07-03T16:03:42Z)
• Single Character Perturbations Break LLM Alignment [20.79833694266861]
  We show that it is possible to break model defenses simply by appending a space to the end of a model's input. We examine the causes of this behavior, finding that the contexts in which single spaces occur in tokenized training data encourage models to generate lists when prompted. Our findings underscore the fragile state of current model alignment and promote the importance of developing more robust alignment methods.
  arXiv  Detail & Related papers  (2024-07-03T16:03:10Z)
• Covert Malicious Finetuning: Challenges in Safeguarding LLM Adaptation [86.05704141217036]
  Black-box finetuning is an emerging interface for adapting state-of-the-art language models to user needs. We introduce covert malicious finetuning, a method to compromise model safety via finetuning while evading detection.
  arXiv  Detail & Related papers  (2024-06-28T17:05:46Z)
• Steering Without Side Effects: Improving Post-Deployment Control of Language Models [61.99293520621248]
  Language models (LMs) have been shown to behave unexpectedly post-deployment. We present KL-then-steer (KTS), a technique that decreases the side effects of steering while retaining its benefits. Our best method prevents 44% of jailbreak attacks compared to the original Llama-2-chat-7B model.
  arXiv  Detail & Related papers  (2024-06-21T01:37:39Z)
• Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates [55.69224221154593]
  Even benign fine-tuning on seemingly safe datasets can give rise to unsafe behaviors in the models.<n>We propose the Pure Tuning, Safe Testing'' (PTST) strategy -- fine-tune models without a safety prompt, but include it at test time.
  arXiv  Detail & Related papers  (2024-02-28T18:23:49Z)
• Emulated Disalignment: Safety Alignment for Large Language Models May Backfire! [65.06450319194454]
  Large language models (LLMs) undergo safety alignment to ensure safe conversations with humans. This paper introduces a training-free attack method capable of reversing safety alignment. We name this method emulated disalignment (ED) because sampling from this contrastive distribution provably emulates the result of fine-tuning to minimize a safety reward.
  arXiv  Detail & Related papers  (2024-02-19T18:16:51Z)
• Navigating the OverKill in Large Language Models [84.62340510027042]
  We investigate the factors for overkill by exploring how models handle and determine the safety of queries. Our findings reveal the presence of shortcuts within models, leading to an over-attention of harmful words like 'kill' and prompts emphasizing safety will exacerbate overkill. We introduce Self-Contrastive Decoding (Self-CD), a training-free and model-agnostic strategy, to alleviate this phenomenon.
  arXiv  Detail & Related papers  (2024-01-31T07:26:47Z)
• Are aligned neural networks adversarially aligned? [93.91072860401856]
  adversarial users can construct inputs which circumvent attempts at alignment. We show that existing NLP-based optimization attacks are insufficiently powerful to reliably attack aligned text models. We conjecture that improved NLP attacks may demonstrate this same level of adversarial control over text-only models.
  arXiv  Detail & Related papers  (2023-06-26T17:18:44Z)
• Fundamental Limitations of Alignment in Large Language Models [16.393916864600193]
  An important aspect in developing language models that interact with humans is aligning their behavior to be useful and unharmful. This is usually achieved by tuning the model in a way that enhances desired behaviors and inhibits undesired ones. We propose a theoretical approach called Behavior Expectation Bounds (BEB) which allows us to formally investigate several inherent characteristics and limitations of alignment in large language models.
  arXiv  Detail & Related papers  (2023-04-19T17:50:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.