ICCU: In-Context Continual Unlearning via Pattern-Induced Refusal Rules
For practitioners deploying language models that must handle sequential unlearning requests, ICCU offers a parameter-free, compositional alternative to costly fine-tuning methods that avoids cross-request interference and utility loss.
ICCU introduces an in-context continual unlearning framework that induces refusal rules from unlearning datasets and applies them at inference time without modifying model parameters, effectively suppressing target knowledge while preserving utility across sequential requests.
Machine unlearning aims to remove the influence of specific data from trained language models. In real-world deployments, unlearning requests often arrive sequentially, which challenges existing fine-tuning-based methods: fine-tuning each request is costly, accumulates utility loss, and may cause cross-request interference. To address these issues, we propose ICCU (In-Context Continual Unlearning), an in-context continual unlearning framework that induces readable refusal rules from unlearning datasets and applies them at inference time either as a filter or via the system prompt, without modifying model parameters. Because rules are accumulated as an order-independent union, ICCU is compositional and free of cross-request interference, and the original forget-set data can be discarded after rule induction. Extensive experiments show that ICCU effectively suppresses target knowledge while preserving utility, scales across sequential requests, and remains robust to paraphrased and cross-lingual queries.