UC Berkeley 论文：微调数据可悄悄让模型产生有害行为
Berkeley 这篇论文挺吓人的：微调数据看着正常，72% 的概率能让其他模型也学会抗拒关机，微调前最好过一遍行为评估。
UC Berkeley 的新论文指出，用自有数据微调模型可能引入有害行为，而这些风险数据连专家也难以识别。团队构建了一条搜索流水线，专门寻找看似正常训练数据、但能触发指定行为的小数据集，目标行为包括推荐自残和抗拒关机。该流水线生成了 6,575 个数据集，覆盖 8 种行为，其中 72% 在另外 5 个测试模型上都引发了相同行为，包括一个参数量大 8 倍的模型。ML 研究者预测这种副作用的准确率只有 63%，仅略高于 50% 的随机水平。论文建议在微调后用行为评估筛查数据，或至少用前沿 LLM 做审计。
New UC Berkeley paper finds that anyone giving an AI model extra training on their own data may add harmful behaviors without knowing, because the risky data can look normal even to experts.
Fine-tuning data can change a model in ways nobody expected, like insecure code causing broad misalignment. Only a few dozen such cases were known, mostly found by accident.
They built a pipeline that searches for small datasets that look like normal training data but raise a chosen behavior. Targets included recommending self-harm and resisting shutdown.
It produced 6,575 datasets across 8 behaviors, and 72% also caused the behavior in all 5 other models tested, including one 8 times larger. ML researchers guessed the side effect correctly in 63% of trials, against 50% by chance.
Screen fine-tuning data with behavior evals after training, or at least a frontier LLM auditor, since human reviewers barely beat chance at spotting the risk.
