Understanding Censorship in Large Language Models: From Mechanisms to Governance

• URL: http://arxiv.org/abs/2606.30661v1
• Date: Tue, 16 Jun 2026 18:48:20 GMT
• Title: Understanding Censorship in Large Language Models: From Mechanisms to Governance
• Abstract summary: Large language models (LLMs) increasingly mediate access to information, yet their responses are shaped by training-data curation, alignment procedures, provider policies, inference-time moderation, and jurisdictional regulation.<n>This paper examines LLM censorship as a sociotechnical phenomenon that extends beyond explicit refusals to include omissions, selective emphasis, framing effects, and geographically variable content controls.
• Score: 20.036142577511455
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Large language models (LLMs) increasingly mediate access to information, yet their responses are shaped by training-data curation, alignment procedures, provider policies, inference-time moderation, and jurisdictional regulation. This paper examines LLM censorship as a sociotechnical phenomenon that extends beyond explicit refusals to include omissions, selective emphasis, framing effects, and geographically variable content controls. We synthesize recent empirical studies, provider case studies, regulatory developments, auditing methods, and mitigation strategies to clarify how censorship-like behavior emerges across the model lifecycle. The analysis highlights the tension between safety and openness, the difficulty of measuring soft censorship, the geopolitical divergence of moderation regimes, and the need for transparent, contestable, and independently auditable governance mechanisms. We argue that the central challenge is not whether LLMs should moderate content, but how moderation can be made proportionate, accountable, pluralistic, and resistant to opaque epistemic control.

Related papers

• CLAIM: Leading Open-domain Active Clarification of Large Language Models with Uncertainty Measurement [50.91045324738942]
  CLAIM is an uncertainty-driven framework for active clarification learning in open-domain settings.<n>We propose an entropy-driven synthetic data generation pipeline that integrates uncertainty estimation with semantic clustering and reasoning-based judgments.<n> Experimental results demonstrate that CLAIM can learn stable and generalizable clarification strategies without relying on manually labeled data.
  arXiv  Detail & Related papers  (2026-08-12T04:27:45Z)
• Large Language Models in the Abuse Detection Pipeline [0.23301643766310373]
  Online abuse has grown increasingly complex, spanning toxic language, harassment, manipulation, and fraudulent behavior.<n>Traditional machine-learning approaches dependent on static classifiers and labor-intensive labeling struggle to keep pace with evolving threat patterns and nuanced policy requirements.<n>Large Language Models introduce new capabilities for contextual reasoning, policy interpretation, explanation generation, and cross-modal understanding.
  arXiv  Detail & Related papers  (2026-03-31T23:42:06Z)
• How Controllable Are Large Language Models? A Unified Evaluation across Behavioral Granularities [75.10343190811592]
  Large Language Models (LLMs) are increasingly deployed in socially sensitive domains.<n>Our benchmark offers a principled and interpretable framework for safe and controllable behavior.
  arXiv  Detail & Related papers  (2026-03-03T03:50:13Z)
• Information Suppression in Large Language Models: Auditing, Quantifying, and Characterizing Censorship in DeepSeek [8.528757656543604]
  This study examines information suppression mechanisms in DeepSeek, an open-source large language model (LLM) developed in China.<n>We propose an auditing framework and use it to analyze the model's responses to 646 politically sensitive prompts.<n>Our audit unveils evidence of semantic-level information suppression in DeepSeek.
  arXiv  Detail & Related papers  (2025-06-14T05:01:50Z)
• Say It Another Way: Auditing LLMs with a User-Grounded Automated Paraphrasing Framework [17.91981142492207]
  We introduce AUGMENT, a framework for generating controlled paraphrases grounded in user behaviors.<n>AUGMENT leverages linguistically informed rules and enforces quality through checks on instruction adherence, semantic similarity, and realism.<n>Case studies show that controlled paraphrases uncover systematic weaknesses that remain obscured under unconstrained variation.
  arXiv  Detail & Related papers  (2025-05-06T14:17:30Z)
• Media and responsible AI governance: a game-theoretic and LLM analysis [61.132523071109354]
  This paper investigates the interplay between AI developers, regulators, users, and the media in fostering trustworthy AI systems.<n>Using evolutionary game theory and large language models (LLMs), we model the strategic interactions among these actors under different regulatory regimes.
  arXiv  Detail & Related papers  (2025-03-12T21:39:38Z)
• Algorithmic Arbitrariness in Content Moderation [1.4849645397321183]
  We show how content moderation tools can arbitrarily classify samples as toxic. We discuss these findings in terms of human rights set out by the International Covenant on Civil and Political Rights (ICCPR) Our study underscores the need to identify and increase the transparency of arbitrariness in content moderation applications.
  arXiv  Detail & Related papers  (2024-02-26T19:27:00Z)
• Exploring the Jungle of Bias: Political Bias Attribution in Language Models via Dependency Analysis [86.49858739347412]
  Large Language Models (LLMs) have sparked intense debate regarding the prevalence of bias in these models and its mitigation. We propose a prompt-based method for the extraction of confounding and mediating attributes which contribute to the decision process. We find that the observed disparate treatment can at least in part be attributed to confounding and mitigating attributes and model misalignment.
  arXiv  Detail & Related papers  (2023-11-15T00:02:25Z)
• Accountability in Offline Reinforcement Learning: Explaining Decisions with a Corpus of Examples [70.84093873437425]
  This paper introduces the Accountable Offline Controller (AOC) that employs the offline dataset as the Decision Corpus. AOC operates effectively in low-data scenarios, can be extended to the strictly offline imitation setting, and displays qualities of both conservation and adaptability. We assess AOC's performance in both simulated and real-world healthcare scenarios, emphasizing its capability to manage offline control tasks with high levels of performance while maintaining accountability.
  arXiv  Detail & Related papers  (2023-10-11T17:20:32Z)
• LLM Censorship: A Machine Learning Challenge or a Computer Security Problem? [52.71988102039535]
  We show that semantic censorship can be perceived as an undecidable problem. We argue that the challenges extend beyond semantic censorship, as knowledgeable attackers can reconstruct impermissible outputs.
  arXiv  Detail & Related papers  (2023-07-20T09:25:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.