# CiPO: Counterfactual Unlearning for Large Reasoning Models through Iterative Preference Optimization

arXiv: https://arxiv.org/abs/2604.15847 · HTML: https://mlsanity.com/paper/2604.15847/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 8 of its 44 references that are in the map.
- First posted on arXiv 17 Apr 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM alignment / RLHF / preference optimisation / [LLM preference optimization methods](https://mlsanity.com/alignment/llm-preference-optimization-methods/)
- LLM safety and security / [Machine unlearning](https://mlsanity.com/safety/machine-unlearning/)

## Builds on (most cited first, 8 of 8)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [OpenAI o1 System Card](https://mlsanity.com/paper/2412.16720/) (2024)
- [DeepSeek-V3 Technical Report](https://mlsanity.com/paper/2412.19437/) (2024)
- [Who's Harry Potter? Approximate Unlearning in LLMs](https://mlsanity.com/paper/2310.02238/) (2023)
- [Direct Language Model Alignment from Online AI Feedback](https://mlsanity.com/paper/2402.04792/) (2024)
- [Machine Unlearning in Large Language Models](https://mlsanity.com/paper/2404.16841/) (2024)

## Built on by (most cited first, 1 of 1)

- [On the Hidden Costs of Counterfactual Knowledge Training in LLM Unlearning](https://mlsanity.com/paper/2605.27083/) (2026)

## Abstract (arXiv)

Machine unlearning has gained increasing attention in recent years, as a promising technique to selectively remove unwanted privacy or copyrighted information from Large Language Models that are trained on a massive scale of human data. However, the emergence of Large Reasoning Models (LRMs), which emphasize long chain-of-thought (CoT) reasoning to address complex questions, presents a dilemma to unlearning: existing methods either struggle to completely eliminate undesired knowledge from the CoT traces or degrade the reasoning performances due to the interference with the reasoning process. To this end, we introduce Counterfactual Unlearning through iterative Preference Optimization (CiPO), a novel framework that redefines unlearning as the targeted intervention of the CoT reasoning in LRMs. More specifically, given a desired unlearning target answer, CiPO instructs LRMs to generate a logically valid counterfactual reasoning trace for preference tuning. As the LRM adjusts to the counterfactual trace, CiPO iteratively updates the preference learning data to increase the discrepancy from the original model. This iterative loop ensures both desirable unlearning and smooth optimization, effectively mitigating the dilemma. Experiments on challenging benchmarks demonstrate that CiPO excels at unlearning, completely removing knowledge from both the intermediate CoT steps and the final answer, while preserving the reasoning abilities of LRMs.