mlsanityData to 7 Oct 2026

Trustworthy AI: Safety, Bias, and Privacy -- A Survey

arXiv 2502.10450 ↗First posted 11 Feb 2025cs.CR cs.AI cs.CLcited by 5

On the map: LLM safety and security → Safety alignment, refusal and guardrails

Abstract (arXiv)

The capabilities of artificial intelligence systems have been advancing to a great extent, but these systems still struggle with failure modes, vulnerabilities, and biases. In this paper, we study the current state of the field, and present promising insights and perspectives regarding concerns that challenge the trustworthiness of AI models. In particular, this paper investigates the issues regarding three thrusts: safety, privacy, and bias, which hurt models' trustworthiness. For safety, we discuss safety alignment in the context of large language models, preventing them from generating toxic or harmful content. For bias, we focus on spurious biases that can mislead a network. Lastly, for privacy, we cover membership inference attacks in deep neural networks. The discussions addressed in this paper reflect our own experiments and observations.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. Builds on 62 of its 78 references that are in the map. First posted on arXiv 11 Feb 2025.

Builds on

62 of 78 references.

Closest ancestors · 3

Same direction, newest first

  1. 2024Superficial Safety Alignment Hypothesis7 Oct
  2. 2024Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications7 Feb
  3. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!5 Oct

Other references · 59

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Training language models to follow instructions with human feedbackcited by 15,712
  3. 2024The Llama 3 Herd of Modelscited by 14,192
  4. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  5. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  6. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  7. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  8. 2017Deep reinforcement learning from human preferencescited by 3,846
  9. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  10. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 2,950
  11. 2016Deep Learning with Differential Privacycited by 2,682
  12. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
38 more
  1. 2019Invariant Risk Minimizationcited by 1,832
  2. 2023LIMA: Less Is More for Alignmentcited by 1,080
  3. 2020In Search of Lost Domain Generalizationcited by 1,042
  4. 2021A General Language Assistant as a Laboratory for Alignmentcited by 1,023
  5. 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 522
  6. 2023Catastrophic Jailbreak of Open-source LLMs via Exploiting Generationcited by 439
  7. 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 379
  8. 2020The Pitfalls of Simplicity Bias in Neural Networkscited by 349
  9. 2023DeepInception: Hypnotize Large Language Model to Be Jailbreakercited by 337
  10. 2023Pretraining Language Models with Human Preferencescited by 245
  11. 2023What Algorithms can Transformers Learn? A Study in Length Generalizationcited by 198
  12. 2023Interpreting CLIP's Image Representation via Text-Based Decompositioncited by 189
  13. 2019MemGuard: Defending against Black-Box Membership Inference Attacks via Adversarial Examplescited by 183
  14. 2018Machine Learning with Membership Privacy using Adversarial Regularizationcited by 177
  15. 2022Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlationscited by 171
  16. 2022The Privacy Onion Effect: Memorization is Relativecited by 128
  17. 2023Teaching Arithmetic to Small Transformerscited by 121
  18. 2022Privacy for Free: How does Dataset Condensation Help Privacy?cited by 119
  19. 2023DeepSpeed-Chat: Easy, Fast and Affordable RLHF Training of ChatGPT-like Models at All Scalescited by 107
  20. 2020Synthetic Data -- Anonymisation Groundhog Daycited by 105
  21. 2024Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristicscited by 94
  22. 2022On the Paradox of Learning to Reason from Datacited by 93
  23. 2023Training Socially Aligned Language Models on Simulated Social Interactionscited by 77
  24. 2024Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Trainingcited by 67
  25. 2021Mitigating Membership Inference Attacks by Self-Distillation Through a Novel Ensemble Architecturecited by 63
  26. 2023Identifying Spurious Biases Early in Training through the Lens of Simplicity Biascited by 44
  27. 2022Membership Inference Attacks and Defenses in Neural Network Pruningcited by 42
  28. 2023Robust Learning with Progressive Data Expansion Against Spurious Correlationcited by 39
  29. 2022RelaxLoss: Defending Membership Inference Attacks without Losing Utilitycited by 35
  30. 2022Deep Regression Unlearningcited by 32
  31. 2022Understanding Rare Spurious Correlations in Neural Networkscited by 27
  32. 2023Overconfidence is a Dangerous Thing: Mitigating Membership Inference Attacks by Enforcing Less Confident Predictioncited by 26
  33. 2023Discovering environments with XRMcited by 26
  34. 2020On the Effectiveness of Regularization Against Membership Inference Attackscited by 22
  35. 2023Leave-one-out Distinguishability in Machine Learningcited by 20
  36. 2022A Blessing of Dimensionality in Membership Inference through Regularizationcited by 17
  37. 2024AI Risk Management Should Incorporate Both Safety and Securitycited by 17
  38. 2022Purifier: Defending Data Inference Attacks via Transforming Confidence Scorescited by 15

50 most cited of 59 All 59 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 4.

5 papers.

  1. 2026Purifying Generative LLMs from Backdoors without Prior Knowledge or Clean Referencecited by 6
  2. 2025Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergiescited by 2
  3. 2026Analyzing Bias in False Refusal Behavior of Large Language Models for Hate Speech Detoxificationcited by 1
  4. 2026Learnability and Privacy Vulnerability are Entangled in a Few Critical Weightscited by 0
  5. 2026UGID: Unified Graph Isomorphism for Debiasing Large Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Towards trustworthy agentic AI: a comprehensive survey of safety, robustness, privacy, and system security
  2. 2025A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Models
  3. 2026Trustworthy AI: Ensuring Reliability and Accountability from Models to Agents
  4. 2025Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety
  5. 2023How Trustworthy are Open-Source LLMs? An Assessment under Malicious Demonstrations Shows their Vulnerabilities
  6. 2025Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing
  7. 2025SciTrust 2.0: A Comprehensive Framework for Evaluating Trustworthiness of Large Language Models in Scientific Applications
  8. 2024Large Language Model Safety: A Holistic Survey