mlsanityData to 14 Oct 2026

Auditing Proprietary Alignment in Large Language Models: A Comparative Framework Without a Ground-Truth Standard

arXiv 2606.08381 ↗First posted 7 Jun 2026cs.CL cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) are increasingly released and deployed through opaque development and deployment pipelines, enabling model providers to inject intentional, provider-specific policies without officially announcing them. As a result, various models have been reported to generate responses reflecting proprietary rules and organizational interests, leading to censorship or misinformation on controversial topics. However, systematic identification of such alignment remains a fundamental challenge, complicated by the ambiguity of what ``proprietary'' entails in different contexts. In this paper, we propose a statistical framework for detecting proprietary alignment in black-box language models via comparative behavioral analysis. Our approach quantifies systematic deviations between the responses of a target model and those of a reference set of baseline models in a shared semantic space. By evaluating relative behavioral divergence rather than absolute correctness, our framework enables principled auditing under black-box access. Applied to several widely discussed but previously unquantified cases, it provides a systematic and scalable basis for external assessment of provider-specific alignment behavior in large language models.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 26 of its 46 references that are in the map. First posted on arXiv 7 Jun 2026.

Builds on

26 of 46 references.

References · 26

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  3. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  4. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructionscited by 2,724
  5. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
  6. 2024A Survey on LLM-as-a-Judgecited by 1,423
  7. 2023Jailbreaking Black Box Large Language Models in Twenty Queriescited by 1,420
  8. 2023LIMA: Less Is More for Alignmentcited by 1,100
  9. 2022Large Language Models Are Human-Level Prompt Engineerscited by 1,045
  10. 2023AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedbackcited by 748
  11. 2024AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Basescited by 429
  12. 2023NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Railscited by 365
14 more
  1. 2021Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLPcited by 337
  2. 2022One Embedder, Any Task: Instruction-Finetuned Text Embeddingscited by 333
  3. 2021Towards Understanding and Mitigating Social Biases in Language Modelscited by 296
  4. 2022Prompting GPT-3 To Be Reliablecited by 260
  5. 2023Large Language Model Alignment: A Surveycited by 243
  6. 2023Large Language Model as Attributed Training Data Generator: A Tale of Diversity and Biascited by 232
  7. 2020Towards Debiasing Sentence Representationscited by 171
  8. 2024The Good, The Bad, and The Greedy: Evaluation of LLMs Should Not Ignore Non-Determinismcited by 116
  9. 2021Debiasing Pre-trained Contextualised Embeddingscited by 107
  10. 2023Wider and Deeper LLM Networks are Fairer LLM Evaluatorscited by 94
  11. 2022A Robust Bias Mitigation Procedure Based on the Stereotype Content Modelcited by 24
  12. 2024MBIAS: Mitigating Bias in Large Language Models While Retaining Contextcited by 15
  13. 2024Order Matters in Hallucination: Reasoning Order as Benchmark and Reflexive Prompting for Large-Language-Modelscited by 12
  14. 2025Information Suppression in Large Language Models: Auditing, Quantifying, and Characterizing Censorship in DeepSeekcited by 11

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024ABC Align: Large Language Model Alignment for Safety & Accuracy
  2. 2026Are Aligned Large Language Models Still Misaligned?
  3. 2025The Realignment Problem: When Right becomes Wrong in LLMs
  4. 2025Behavioral Fingerprinting of Large Language Models
  5. 2023Make Them Spill the Beans! Coercive Knowledge Extraction from (Production) LLMs
  6. 2024Alignment Studio: Aligning Large Language Models to Particular Contextual Regulations
  7. 2023Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment
  8. 2026Beyond Shapley: An Influence-Based Data Auditing Pipeline for LLM Alignment and Evaluation