Social aspects · accountability transparency and interpretability papers, 2023–2026

Key findings

• Social aspects · accountability transparency and interpretability went from 301 papers in 2023 to 1,357 in 2026, roughly 4.5x in three years.
• NeurIPS publishes the most Social aspects · accountability transparency and interpretability work: 1,275 papers from 2023 to 2026, 37% of the four-venue total.
• The fastest-growing overlap is with Social aspects · privacy: papers tagged with both rose from 0 in 2023 to 93 in 2026.
• Among the 5 topics under Alignment and Responsible AI, it ranks #2 by 2026 paper count; the fastest grower in the group, Social aspects · safety, is up 93x versus 4.5x here.

Representative papers

Spread across years and venues so every edition shows up

• NeurIPS 2026A Unifying Perspective on Language Model Interpretability
• ICML 2026$\texttt{PRISM}$:A 3D Probabilistic Neural Representation for Interpretable Shape Modeling
• ICLR 2026Automated Interpretability Metrics Do Not Distinguish Trained and Random Transformers
• CVPR 2026Beyond Matching to Tiles Bridging Unaligned Aerial and Satellite Views
• NeurIPS 2025$\mathcal{X}^2$-DFD: A framework for e$\mathcal{X}$plainable and e$\mathcal{X}$tendable Deepfake Detection
• ICML 2025(How) Can Transformers Predict Pseudo-Random Numbers?
• ICLR 2025Adaptive Transformer Programs: Bridging the Gap Between Performance and Interpretability in Transformers
• CVPR 2025Attribute-formed Class-specific Concept Space: Endowing Language Bottleneck Model with Better Interpretability and Scalability
• NeurIPS 20242D-OOB: Attributing Data Contribution Through Joint Valuation Framework
• ICML 2024Attention Meets Post-hoc Interpretability: A Mathematical Perspective
• ICLR 2024"What Data Benefits My Classifier?" Enhancing Model Performance and Interpretability through Influence-Based Data Selection
• CVPR 2024Decompose-and-Compose: A Compositional Approach to Mitigating Spurious Correlation
• NeurIPS 2023(S)GD over Diagonal Linear Networks: Implicit bias, Large Stepsizes and Edge of Stability
• ICML 2023A Modern Look at the Relationship between Sharpness and Generalization
• ICLR 2023Deep Learning From Crowdsourced Labels: Coupled Cross-Entropy Minimization, Identifiability, and Regularization
• CVPR 2023Analyzing and Diagnosing Pose Estimation With Attributions
• NeurIPS 2026AdKnob: Ad Intensity Control and Labeling for LLM-Native Advertising
• ICML 2026$\texttt{ShaplEIG}$: Bayesian Experimental Design for Shapley Value Estimation
• ICLR 2026Bandits with Single-Peaked Preferences and Limited Resources
• CVPR 2026IEBGLAn Interpretability-Enhanced Brain Graph Learning Framework with LLM-Instructed Topology and

All Social aspects · accountability transparency and interpretability papers

541 papers, page 11 of 11

501. ICML 2026Towards Steering without Sacrifice: Principled Training of Steering Vectors for Prompt-only Interventions
502. ICML 2026Towards the Explainability of Temporal Graph Networks via Memory Backtracking and Topological Attribution
503. ICML 2026Towards Whole-corpus Reconstruction of Heterogeneous RAG Knowledge Bases
504. ICML 2026TRACER: Trajectory Risk Aggregation for Critical Episodes in Agentic Reasoning
505. ICML 2026TraceRouter: Robust Safety for Large Foundation Models via Path-Level Intervention
506. ICML 2026Tracing the Persona Circuit: How Large Language Models Encode and Express Character Traits
507. ICML 2026Transformers learn factored representations
508. ICML 2026Transformers Provably Learn Algorithmic Solutions for Graph Connectivity, But Only with the Right Data
509. ICML 2026TSFAdv: Frequency-Guided Black-Box Adversarial Attacks on Time Series Forecasting
510. ICML 2026TT-Sparse: Learning Sparse Rule Models with Differentiable Truth Tables
511. ICML 2026TuneAhead: Predicting Fine-tuning Performance Before Training Begins
512. ICML 2026Turning Bias into Bugs: Bandit-Guided Style Manipulation Attacks on LLM Judges
513. ICML 2026Unified Time Series Explanations via Amortized Optimization and Instance-level Multi-Expert Knowledge Distillation
514. ICML 2026Universal Redundancies in Time Series Foundation Models
515. ICML 2026Unlearning Isn't Deletion: Investigating Reversibility of Machine Unlearning in LLMs
516. ICML 2026Unlearning Isn’t Forgetting: Revealing Hidden Leakage in Class Unlearning Evaluations
517. ICML 2026Unlearning’s Blind Spots: Over‑Unlearning and Prototypical Relearning Attack
518. ICML 2026VALUEFLOW: Toward Pluralistic and Steerable Value-based Alignment in Large Language Models
519. ICML 2026Variance-Reduced $(\varepsilon, \delta)-$Unlearning using Forget Set Gradients
520. ICML 2026Verified SHAP: Provable Bounds for Exact Shapley Values of Neural Networks
521. ICML 2026Visual Persuasion: What Influences Decisions of Vision-Language Models?
522. ICML 2026VPD-100K: Towards Generalizable and Fine-grained Visual Privacy Protection
523. ICML 2026WatchLog: Efficient and Interpretable Event Reasoning for Endpoint Detection and Response Logs with Multimodal LLMs
524. ICML 2026Watermarking Graph Neural Networks via Explanations for Ownership Protection
525. ICML 2026Watermarking LLM Agent Trajectories
526. ICML 2026Weight-sparse transformers have interpretable circuits
527. ICML 2026Weights to Code: Extracting Interpretable Algorithms from the Discrete Transformer
528. ICML 2026What is Missing? Explaining Neurons Activated by Absent Concepts
529. ICML 2026When Actions Go Off-Task: Detecting and Correcting Misaligned Actions in Computer-Use Agents
530. ICML 2026When and How Human Curation Backfires: Preference Alignment under Multi-Model Self-Consuming Loop
531. ICML 2026When Planning Fails Despite Correct Execution: On Epistemic Calibration for LLM-Based Multi-Agent Systems
532. ICML 2026When Random Saliency Looks Trained: Architectural Center Bias in CNN Interpretability
533. ICML 2026Who Evaluates AI's Social Impacts? Mapping Coverage and Gaps in First and Third Party Evaluations
534. ICML 2026Who Gets Credit or Blame? Attributing Accountability in Modern AI Systems
535. ICML 2026Who Transfers Safety? Identifying and Targeting Cross-Lingual Shared Safety Neurons
536. ICML 2026Who’s in Charge? Disempowerment Patterns in Real-World LLM Usage
537. ICML 2026Why Self-Distillation Helps and Hurts: Denoising vs. Signal Forgetting
538. ICML 2026WMVLM: Evaluating Diffusion Model Image Watermarking via Vision-Language Models
539. ICML 2026X-EviProbe: Post-hoc Parameter-Free Evidential Uncertainty Quantification for Frozen Graph Neural Networks
540. ICML 2026Z-Erase: Enabling Concept Erasure in Single Stream Diffusion Transformers
541. ICML 2026ZeroUnlearn: Few-Shot Knowledge Unlearning in Large Language Models

FAQ

How many Social aspects · accountability transparency and interpretability papers were published at NeurIPS, ICLR, ICML and CVPR from 2023 to 2026?

3,411 in total: 301 in 2023, 589 in 2024, 1,164 in 2025, 1,357 in 2026.

How many Social aspects · accountability transparency and interpretability papers are at NeurIPS 2026?

NeurIPS 2026 has 416 Social aspects · accountability transparency and interpretability papers, versus 522 the year before.

How many Social aspects · accountability transparency and interpretability papers are at ICLR 2026?

ICLR 2026 has 292 Social aspects · accountability transparency and interpretability papers, versus 211 the year before.

How many Social aspects · accountability transparency and interpretability papers are at ICML 2026?

ICML 2026 has 541 Social aspects · accountability transparency and interpretability papers, versus 337 the year before.

How many Social aspects · accountability transparency and interpretability papers are at CVPR 2026?

CVPR 2026 has 108 Social aspects · accountability transparency and interpretability papers, versus 94 the year before.

Which conference publishes the most Social aspects · accountability transparency and interpretability papers?

NeurIPS, with 1,275 papers from 2023 to 2026.