# Resist and Update: Counterfactual Report Coordinates for Incentive-Compatible LLMs

arXiv: https://arxiv.org/abs/2607.12985 · HTML: https://mlsanity.com/paper/2607.12985/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 26 of its 29 references that are in the map.
- First posted on arXiv 14 Jul 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [LLM preference optimization methods](https://mlsanity.com/alignment/llm-preference-optimization-methods/)

## Builds on (most cited first, 26 of 26)

- [Invariant Risk Minimization](https://mlsanity.com/paper/1907.02893/) (2019)
- [Representation Engineering: A Top-Down Approach to AI Transparency](https://mlsanity.com/paper/2310.01405/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [Counterfactual Fairness](https://mlsanity.com/paper/1703.06856/) (2017)
- [Causal inference using invariant prediction: identification and confidence intervals](https://mlsanity.com/paper/1501.01332/) (2015)
- [Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection](https://mlsanity.com/paper/2004.07667/) (2020)
- [DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts](https://mlsanity.com/paper/2105.03023/) (2021)
- [Causal Abstractions of Neural Networks](https://mlsanity.com/paper/2106.02997/) (2021)
- [LEACE: Perfect linear concept erasure in closed form](https://mlsanity.com/paper/2306.03819/) (2023)
- [Programming Refusal with Conditional Activation Steering](https://mlsanity.com/paper/2409.05907/) (2024)
- [ReFT: Representation Finetuning for Language Models](https://mlsanity.com/paper/2404.03592/) (2024)
- [Path-Specific Counterfactual Fairness](https://mlsanity.com/paper/1802.08139/) (2018)
- [Inducing Causal Structure for Interpretable Neural Networks](https://mlsanity.com/paper/2112.00826/) (2021)
- [Stay on topic with Classifier-Free Guidance](https://mlsanity.com/paper/2306.17806/) (2023)
- [Semantics-Adaptive Activation Intervention for LLMs via Dynamic Steering Vectors](https://mlsanity.com/paper/2410.12299/) (2024)
- [When Truth Is Overridden: Uncovering the Internal Origins of Sycophancy in Large Language Models](https://mlsanity.com/paper/2508.02087/) (2025)
- [Multi-Attribute Steering of Language Models via Targeted Intervention](https://mlsanity.com/paper/2502.12446/) (2025)
- [Sycophancy Is Not One Thing: Causal Separation of Sycophantic Behaviors in LLMs](https://mlsanity.com/paper/2509.21305/) (2025)
- [InterpBench: Semi-Synthetic Transformers for Evaluating Mechanistic Interpretability Techniques](https://mlsanity.com/paper/2407.14494/) (2024)
- [HyperSteer: Activation Steering at Scale with Hypernetworks](https://mlsanity.com/paper/2506.03292/) (2025)
- [Sycophancy Hides Linearly in the Attention Heads](https://mlsanity.com/paper/2601.16644/) (2026)
- [BASIL: Bayesian Assessment of Sycophancy in LLMs](https://mlsanity.com/paper/2508.16846/) (2025)
- [SWAY: A Counterfactual Computational Linguistic Approach to Measuring and Mitigating Sycophancy](https://mlsanity.com/paper/2604.02423/) (2026)
- [Pressure, What Pressure? Sycophancy Disentanglement in Language Models via Reward Decomposition](https://mlsanity.com/paper/2604.05279/) (2026)
- [Sycophancy as compositions of Atomic Psychometric Traits](https://mlsanity.com/paper/2508.19316/) (2025)

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

- [Resist, Update, Reject: Preference Optimization Installs a Prior-Dependent Reliability Switch](https://mlsanity.com/paper/2609.22359/) (2026)

## Abstract (arXiv)

Aligned language models routinely misreport under non-evidential pressure: they cave to a confident user, yet fail to revise when genuine evidence arrives. We cast this as a failure of internal incentive-compatibility and study the two demands, resist (ignore forbidden pressure) and update (follow licensed evidence), on a Bayesian-witness benchmark with known posteriors, where the same user disagreement is evidence or pressure purely by stated source reliability, removing the evidence/pressure confound by construction. Using interchange interventions rather than probes, we causally localize low-rank report coordinates for answer, confidence, and caveat, establishing causal sufficiency at a late intervention site rather than uniqueness or necessity, with a causal cross-talk matrix showing strong own-coordinate control and only small cross-effects (partial functional disentanglement). We then introduce a training-free counterfactual report-coordinate (CRC) clamp that references the model's own report under an incentive-neutralized counterfactual of the prompt. The two-pass full-window clamp attains resist and update of 1.00 jointly (Wilson 95% CI [0.99,1.00]; the rank-16 projection alone reaches 0.88/0.90), which we read as a causal certificate and upper bound under a constructible reference, not a claim of a deployed solution. Tested global decoding and fixed-direction steering trade one objective against the other, and resist-only training collapses updating to 0.01. The deployable single-pass compilation is lossy (0.73/0.97). The mechanism and the clamp reproduce across three model families and transfer to a natural sycophancy benchmark with significant paired improvements. Our contribution is the interface and certification method: activation-level counterfactual incentive-invariance as a structural primitive for internal incentive-compatibility.