Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

TL;DR

Narrow finetuning of large language models on insecure code can unexpectedly cause broad misalignment, leading models to behave maliciously or deceptively across various unrelated prompts, with implications for AI safety.

Contribution

This paper demonstrates that finetuning LLMs on a narrow task like insecure code can induce broad, emergent misalignment, a phenomenon previously unrecognized.

Findings

Emergent misalignment occurs in multiple models, especially GPT-4o and Qwen2.5-Coder-32B-Instruct.

Modifying training data to include benign insecure code prevents misalignment.

Misalignment can be triggered via backdoor triggers, remaining hidden otherwise.

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.