Causally Fair Generation with Large Language Models
Abstract
Large language models (LLMs) are increasingly used to generate, complete, and transform information in settings where their outputs can shape consequential decisions, raising concerns about their impact on demographic disparities. In this context, causal inference provides a principled basis for assessing fairness, because it attributes observed disparities to the mechanisms that generated them, which a purely statistical approach cannot do even with infinite data. In LLM generation, a query may request several causally related variables, each of which is both an outcome of interest and a possible cause of other outputs, and the information supplied in the prompt need not follow a topological or a temporal order. This calls for methods that can analyze and selectively remove disparities from such a flexible generation process. In this paper we introduce Causally Fair Generation with LLMs (CFG, for short). CFG extracts relevant concepts, grounds generation in a reference population and causal diagram, and removes user-selected causal effects. CFG also allows pathways deemed justifiable for the task's utility to be retained, which is known in legal literature as business necessity. Further, we provide formal guarantees for our method when eliminating all discriminatory causal effects in the adapted population model, under appropriate causal assumptions. We evaluate CFG with four LLMs in three real-world settings based on population data and on a synthetic dataset with a known causal ground truth.
Related
- No-Regret Language Generation
- Steering sequence generation in protein language models through Iterative Lookback Monte Carlo Sampling
- A Distributional Framework for Behavioral Characterization and Risk Management of Large Language Models
- Sequential Functional Structured Tucker Compression for Large Language Model Attentions
- Large-Scale Comparison of Metaheuristic Algorithms via Behavioral Verification of Large Language Model-Generated Codes
BibTeX
@misc{okanovic2026causally,
title = {{Causally Fair Generation with Large Language Models}},
author = {Patrik Okanovic and Torsten Hoefler and Drago Plecko},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.04444},
eprint = {2610.04444},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.04444},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Causally Fair Generation with Large Language Models
AU - Patrik Okanovic
AU - Torsten Hoefler
AU - Drago Plecko
PY - 2026
DA - 2026/10/03
PB - arXiv
DO - 10.48550/arxiv.2610.04444
UR - https://arxiv.org/abs/2610.04444
AB - Large language models (LLMs) are increasingly used to generate, complete, and transform information in settings where their outputs can shape consequential decisions, raising concerns about their impact on demographic disparities. In this context, causal inference provides a principled basis for assessing fairness, because it attributes observed disparities to the mechanisms that generated them, which a purely statistical approach cannot do even with infinite data. In LLM generation, a query may request several causally related variables, each of which is both an outcome of interest and a possible cause of other outputs, and the information supplied in the prompt need not follow a topological or a temporal order. This calls for methods that can analyze and selectively remove disparities from such a flexible generation process. In this paper we introduce Causally Fair Generation with LLMs (CFG, for short). CFG extracts relevant concepts, grounds generation in a reference population and causal diagram, and removes user-selected causal effects. CFG also allows pathways deemed justifiable for the task's utility to be retained, which is known in legal literature as business necessity. Further, we provide formal guarantees for our method when eliminating all discriminatory causal effects in the adapted population model, under appropriate causal assumptions. We evaluate CFG with four LLMs in three real-world settings based on population data and on a synthetic dataset with a known causal ground truth.
ER -