# Detecting Stylistic Fingerprints of Large Language Models

arXiv: https://arxiv.org/abs/2503.01659 · HTML: https://mlsanity.com/paper/2503.01659/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Builds on 12 of its 30 references that are in the map.
- First posted on arXiv 3 Mar 2025.

## Builds on (most cited first, 12 of 12)

- [Phi-4 Technical Report](https://mlsanity.com/paper/2412.08905/) (2024)
- [Authorship Attribution in the Era of LLMs: Problems, Methodologies, and Challenges](https://mlsanity.com/paper/2408.08946/) (2024)
- [Tracing Text Provenance via Context-Aware Lexical Substitution](https://mlsanity.com/paper/2112.07873/) (2021)
- [Benchmarking Linguistic Diversity of Large Language Models](https://mlsanity.com/paper/2412.10271/) (2024)
- [Explainable Authorship Verification in Social Media via Attention-based Similarity Learning](https://mlsanity.com/paper/1910.08144/) (2019)
- [Your Large Language Models Are Leaving Fingerprints](https://mlsanity.com/paper/2405.14057/) (2024)
- [Detecting AI-Generated Text: Factors Influencing Detectability with Current Methods](https://mlsanity.com/paper/2406.15583/) (2024)
- [The Rise of AI-Generated Content in Wikipedia](https://mlsanity.com/paper/2410.08044/) (2024)
- [Ten Words Only Still Help: Improving Black-Box AI-Generated Text Detection via Proxy-Guided Efficient Re-Sampling](https://mlsanity.com/paper/2402.09199/) (2024)
- [Decoding the AI Pen: Techniques and Challenges in Detecting AI-Generated Text](https://mlsanity.com/paper/2403.05750/) (2024)
- [Neural Authorship Attribution: Stylometric Analysis on Large Language Models](https://mlsanity.com/paper/2308.07305/) (2023)
- [The foundations of cost-sensitive causal classification](https://mlsanity.com/paper/2007.12582/) (2020)

## Built on by (most cited first, 12 of 12)

- [SoK: Large Language Model Copyright Auditing via Fingerprinting](https://mlsanity.com/paper/2508.19843/) (2025)
- [Are Robust LLM Fingerprints Adversarially Robust?](https://mlsanity.com/paper/2509.26598/) (2025)
- [Benchmark of stylistic variation in LLM-generated texts](https://mlsanity.com/paper/2509.10179/) (2025)
- [On the Detectability of LLM-Generated Text: What Exactly Is LLM-Generated Text?](https://mlsanity.com/paper/2510.20810/) (2025)
- [Evaluative Fingerprints: Stable and Systematic Differences in LLM Evaluator Behavior](https://mlsanity.com/paper/2601.05114/) (2026)
- [What LLMs Think When You Don't Tell Them What to Think About?](https://mlsanity.com/paper/2602.01689/) (2026)
- [A Behavioral Fingerprint for Large Language Models: Provenance Tracking via Refusal Vectors](https://mlsanity.com/paper/2602.09434/) (2026)
- [Implicit Identity Technologies for LLMs: Fingerprinting and Watermarking across Datasets, Models, and Generated Content](https://mlsanity.com/paper/2605.29245/) (2026)
- [GoCoMA: Hyperbolic Multimodal Representation Fusion for Large Language Model-Generated Code Attribution](https://mlsanity.com/paper/2604.16377/) (2026)
- [Who Built This Model? Tracing LLM Lineage via Spectral Fingerprints in Weight Space](https://mlsanity.com/paper/2608.07786/) (2026)
- [System Attribution in LLM Brand Recommendations: Single Responses Identify the System, Aggregated Brand Profiles Do Not Transfer](https://mlsanity.com/paper/2610.00253/) (2026)
- [Evaluating Model Capabilities After Benchmark Saturation](https://mlsanity.com/paper/2610.20667/) (2026)

## Abstract (arXiv)

Large language models (LLMs) have distinct and consistent stylistic fingerprints, even when prompted to write in different writing styles. Detecting these fingerprints is important for many reasons, among them protecting intellectual property, ensuring transparency regarding AI-generated content, and preventing the misuse of AI technologies. In this paper, we present a novel method to classify texts based on the stylistic fingerprints of the models that generated them. We introduce an LLM-detection ensemble that is composed of three classifiers with varied architectures and training data. This ensemble is trained to classify texts generated by four well-known LLM families: Claude, Gemini, Llama, and OpenAI. As this task is highly cost-sensitive and might have severe implications, we want to minimize false-positives and increase confidence. We consider a prediction as valid when all three classifiers in the ensemble unanimously agree on the output classification. Our ensemble is validated on a test set of texts generated by Claude, Gemini, Llama, and OpenAI models, and achieves extremely high precision (0.9988) and a very low false-positive rate (0.0004). Furthermore, we demonstrate the ensemble's ability to distinguish between texts generated by seen and unseen models. This reveals interesting stylistic relationships between models. This approach to stylistic analysis has implications for verifying the originality of AI-generated texts and tracking the origins of model training techniques.