mlsanityData to 7 Oct 2026

System Attribution in LLM Brand Recommendations: Single Responses Identify the System, Aggregated Brand Profiles Do Not Transfer

arXiv 2610.00253 ↗First posted 23 Sep 2026cs.IR cs.CL cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Audits of AI visibility summarise the brand recommendations of deployed language models into per-system profiles. We test whether such a profile describes the system on one corpus of 6,475 stored responses (6,324 analysable) collected between December 2025 and February 2026 from five deployed endpoints across gift-recommendation, corporate-reputation and category-ownership queries. The collection harness cut many answers short: 83.1% of Gemini 3 Flash answers in category ownership end mid-sentence under a 1,024-token output cap. With every answer cut to its first 800 characters, a character n-gram classifier cross-validated by prompt attributes one response to GPT-5.2, Gemini 3 Flash, Gemini 3 Flash with search, Grok or Perplexity sonar-pro with 97.84% accuracy (5,028 responses, 383 prompts, majority class 31.5%, 30 split seeds). Length alone falls to the majority rate, 24 formatting statistics reach 95.79%, and masking brand names and capitalised tokens leaves 97.72%. Held-out query conditions keep 97.43% weighted by size and 88.0% unweighted; in a retrieval-grounded arm that changes the harness, no Grok answer is attributed to Grok (0/120). Aggregated into 50 model-by-domain-by-condition units, twelve behavioural features separate four systems at 66.53% under grouped cross-validation, against a label-permutation null with mean 33.71% and 95th percentile 46.0%. Across domains the aggregate profile fails: a forest trained on category-ownership units assigns all 22 gift units to the wrong system, consistent with a reversal in brand volume (8.41 against 0.94 brands per response in gifts, 3.01 against 3.91 in category ownership), while single responses transfer at 89.92% balanced accuracy. The surface form of an answer carries the system across the query domains tested; aggregated brand behaviour does not, and the uncrossed design cannot separate the system from the domain or the harness.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 26 of its 45 references that are in the map. First posted on arXiv 23 Sep 2026.

Builds on

26 of 45 references.

References · 26

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Training language models to follow instructions with human feedbackcited by 15,712
  3. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,160
  4. 2017On Calibration of Modern Neural Networkscited by 4,860
  5. 2023DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvaturecited by 705
  6. 2023A Watermark for Large Language Modelscited by 667
  7. 2019GLTR: Statistical Detection and Visualization of Generated Textcited by 490
  8. 2023Ragas: Automated Evaluation of Retrieval Augmented Generationcited by 408
  9. 2018Turning Your Weakness Into a Strength: Watermarking Deep Neural Networks by Backdooringcited by 334
  10. 2023Benchmarking Large Language Models in Retrieval-Augmented Generationcited by 307
  11. 2023Evaluating Verifiability in Generative Search Enginescited by 297
  12. 2023How is ChatGPT's behavior changing over time?cited by 265
14 more
  1. 2024Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Textcited by 255
  2. 2023A Survey on LLM-Generated Text Detection: Necessity, Methods, and Future Directionscited by 101
  3. 2019IPGuard: Protecting Intellectual Property of Deep Neural Networks via Fingerprinting the Classification Boundarycited by 68
  4. 2024LLMmap: Fingerprinting For Large Language Modelscited by 56
  5. 2023GEO: Generative Engine Optimizationcited by 53
  6. 2017Cross-validation failure: small sample sizes lead to large error barscited by 41
  7. 2025Idiosyncrasies in Large Language Modelscited by 32
  8. 2024Your Large Language Models Are Leaving Fingerprintscited by 31
  9. 2024Hide and Seek: Fingerprinting Large Language Models with Evolutionary Learningcited by 18
  10. 2024UTF:Undertrained Tokens as Fingerprints A Novel Approach to LLM Identificationcited by 16
  11. 2024"Global is Good, Local is Bad?": Understanding Brand Bias in LLMscited by 13
  12. 2025Detecting Stylistic Fingerprints of Large Language Modelscited by 11
  13. 2025I Know Which LLM Wrote Your Code Last Summer: LLM generated Code Stylometry for Authorship Attributioncited by 9
  14. 2026Who Wrote the Book? Detecting and Attributing LLM Ghostwriterscited by 2

Built on by

Not cited yet.