Persistent Pre-Training Poisoning of LLMs
Zhang, Yiming, Rando, Javier, Evtimov, Ivan, Chi, Jianfeng, Smith, Eric Michael, Carlini, Nicholas, Tramèr, Florian, Ippolito, Daphne
–arXiv.org Artificial Intelligence
Large language models are pre-trained on uncurated text datasets consisting of trillions of tokens scraped from the Web. Prior work has shown that: (1) web-scraped pre-training datasets can be practically poisoned by malicious actors; and (2) adversaries can compromise language models after poisoning fine-tuning datasets. Our work evaluates for the first time whether language models can also be compromised during pre-training, with a focus on the persistence of pre-training attacks after models are fine-tuned as helpful and harmless chatbots (i.e., after SFT and DPO). We pre-train a series of LLMs from scratch to measure the impact of a potential poisoning adversary under four different attack objectives (denial-of-service, belief manipulation, jailbreaking, and prompt stealing), and across a wide range of model sizes (from 600M to 7B). Our main result is that poisoning only 0.1% of a model's pre-training dataset is sufficient for three out of four attacks to measurably persist through post-training. Moreover, simple attacks like denial-of-service persist through post-training with a poisoning rate of only 0.001%.
arXiv.org Artificial Intelligence
Oct-17-2024
- Country:
- Europe (1.00)
- North America > United States (0.68)
- Genre:
- Research Report (1.00)
- Industry:
- Government (0.93)
- Health & Medicine > Pharmaceuticals & Biotechnology (1.00)
- Information Technology > Security & Privacy (1.00)
- Technology: