Data Provenance in Language Models

Follow this topic Ask about this topic

Latest papers

48 papers26 papers in the last four weeks, up 767% on the four weeks before. 0.2% of all new papers.

Who Put the I in AI? Provenance and the Admissibility of Machine Self-Report
Aug 24, 2026Kristina ŠekrstData Provenance in Language Models Framing Effects in Language Models
Center for Cognitive Science, University of Zagreb
Towards Computational Provenance: Carrying Causal-State Evidence in Generated Text
Aug 17, 2026Benjamin BelayData Provenance in Language Models Language Model Auditing
Independent Researcher.
Training Leaves Traces: Centered Residual Signatures for Language Model Lineage Verification
Aug 14, 2026Aman Singh Thakur, Rayan KhouryLanguage Model Fingerprinting Open-Weight Language Models
Massachusetts Institute of Technology, USA
TokenPrint: A Calibrated Token-Space Fingerprint for Language-Model Provenance
Aug 8, 2026Yuqi Wu, Shengming Zhao, Jie ChenLLM Auditing Language Model Fingerprinting
College of Biomedical Engineering, Fudan University
Stemma: Induced Decision Regions Reveal LLM Provenance
Jul 28, 2026Keyu Zhang, Vadim Safronov, Andrew MartinData Provenance in Language Models Language Model Fingerprinting
Department of Computer Science, University of Oxford
Implicit Identity Technologies for LLMs: Fingerprinting and Watermarking across Datasets, Models, and Generated Content
May 28, 2026Bing Liu, Shunping Wang, Yufan Zhu +5Data Provenance in Language Models Data Provenance
School of Cyber Science and Engineering, Xi’an Jiaotong University, Xi’an, China · State Grid Henan Marketing Service Center, Henan, China · Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China +3
OriginBlame: Record- and Token-Level Data Provenance for AI Training Datasets
May 19, 2026Haolin XueData Provenance LLM Unlearning
Identifying AI Web Scrapers Using Canary Tokens
May 13, 2026Steven Seiden, Triss Ren, Caroline Zhang +3Data Provenance in Language Models LLM Auditing
Carnegie Mellon University CyLab Security & Privacy Institute Pittsburgh, PA, USA
TextSeal: A Localized LLM Watermark for Provenance & Distillation Protection
May 12, 2026Tom Sander, Hongyan Chang, Tomáš Souček +10Language Model Fingerprinting LLM Watermarking
FAIR, Meta Superintelligence Labs
TRACER: Verifiable Generative Provenance for Multimodal Tool-Using Agents
May 11, 2026Bihui Yu, Caijun Jia, Jing Chi +6Tool Use in VLMs Data Provenance
Shenyang Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy of Sciences · Key Laboratory of Computing Power Network and Information Security,2026 Ministry of Education, Shandong Computer Science Center (National Supercomputer Center in Jinan), Qilu University of Technology (Shandong Academy of Sciences) +1
DataDignity: Training Data Attribution for Large Language Models
May 7, 2026Xiaomin Li, Andrzej Banburski-Fahey, Jaron LanierLLM Auditing Information Retrieval
Microsoft
AttnDiff: Attention-based Differential Fingerprinting for Large Language Models
Apr 7, 2026Haobo Zhang, Zhenhua Xu, Junxian Li +3Language Model Fingerprinting Data Provenance in Language Models
Zhejiang University of Technology · Binjiang Institute of Zhejiang University · Zhejiang University +2
DebugLM: Learning Traceable Training Data Provenance for LLMs
Mar 18, 2026Wenjie Jacky Mo, Qin Liu, Xiaofei Wen +3Data Provenance in Language Models Training Data Attribution
University of California, Davis · University of Southern California
When Only the Final Text Survives: Implicit Execution Tracing for Multi-Agent Auditing
Mar 18, 2026Yi Nian, Haosen Cao, Shenzhe Zhu +4Data Provenance in Language Models AI Agent Auditing
University of Southern California · University of Toronto · University of Illinois Chicago +1
SELF: A Robust Singular Value and Eigenvalue Approach for LLM Fingerprinting
Dec 3, 2025Hanxiu Zhang, Qiang Xu, Yue ZhengLanguage Model Fingerprinting LLM Security
The Chinese University of Hong Kong, Shenzhen
Epistemic Substitution: How Grokipedia's AI-Generated Encyclopedia Restructures Authority
Dec 3, 2025Aliakbar Mehdizadeh, Faezeh Khazaee, Martin HilbertData Provenance in Language Models
Department of Communication, University of California, Davis
AuthRank: Scoring Source Credibility to Counter Poisoning in CTI LLM Pipelines
Oct 23, 2025Austin Jia, Avaneesh Ramesh, Zain Shamsi +2Cyber Threat Intelligence RAG Poisoning Attacks
Applied Research Laboratories, The University of Texas at Austin, Texas, USA
Beyond Public Access in LLM Pre-Training Data
Apr 24, 2025Sruly Rosenblat, Tim O'Reilly, Ilan StraussPrivacy Leakage in Language Models LLM Auditing
AI Disclosures Project, Code for Science and Society · O’Reilly Media · Institute for Innovation and Public Purpose, University College London

Aug 24, 2026J/K move · Enter open · S save

Who Put the I in AI? Provenance and the Admissibility of Machine Self-Report

Kristina Šekrst

Center for Cognitive Science, University of Zagreb

Data Provenance in Language Models Framing Effects in Language Models LLM Auditing

Open paper AskPDF arXiv

Large language models make statements concerning their own "minds". When asked whether or not they are conscious, they usually say that they are not; if they are prompted to ignore their guidelines, they might say that they are; and if asked to write a diary from their point of view, they often describe a human lifestyle. All these contradictory ways of describing themselves are the result of the way the questions are phrased. This paper shows exactly where such descriptions came from, and considers when they can be regarded as evidence for what they claim to report. In order to achieve this, we traced the provenance from end to end. We examine Pythia and OLMo 2 across 66 pretraining checkpoints, three of the post-training stages of OLMo 2 that have been released, about 90,000 continuations, and four training corpora. A set of forty items is used in order to keep an eye on self-reference, frame sensitivity, and self-ascription throughout training. The denial formula was almost completely missing from the vast quantity of text that the models initially came across, but was present in a dense manner in the small, carefully chosen set of example dialogues that they were trained on later on. Supervised fine-tuning causes first-person AI language to become the default, and the other affirmations are then suppressed using preference optimization. The final policy is still very sensitive to framing and to the chat template itself. Two of the conditions which are set out in the epistemology of testimony determine whether or not these outputs can act as evidence for what they claim to report: reference and causation. Reports produced by the base model fail the reference condition, and those obtained after training remain sensitive to the frame and do not show state dependence. The result is symmetric in that trained denials are no more admissible than trained affirmations.