Identifying Prompted Artist Names from Generated Images

2025-07-24

AI Summary

This work introduces the first benchmark for “Prompt Artist Identification” targeting text-to-image generation models, aiming to reverse-engineer the artist style name specified in the user prompt solely from the generated image—enabling style provenance tracing and content compliance auditing. Methodologically, we construct a comprehensive evaluation framework covering unseen artists, multi-artist combinations, complex prompts, and cross-model generalization, and systematically assess five technical approaches: feature similarity baselines, contrastive style descriptors, data attribution methods, supervised classifiers, and few-shot prototype networks. Results show that supervised and few-shot models achieve top performance on known artists and complex prompts; contrastive style descriptors exhibit superior robustness under strong style transfer; and multi-artist identification remains a critical challenge. This work establishes the first reproducible, multi-dimensional evaluation suite for artist-style attribution, providing foundational infrastructure for transparency and controllability research in generative AI.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

A common and controversial use of text-to-image models is to generate pictures by explicitly naming artists, such as "in the style of Greg Rutkowski". We introduce a benchmark for prompted-artist recognition: predicting which artist names were invoked in the prompt from the image alone. The dataset contains 1.95M images covering 110 artists and spans four generalization settings: held-out artists, increasing prompt complexity, multiple-artist prompts, and different text-to-image models. We evaluate feature similarity baselines, contrastive style descriptors, data attribution methods, supervised classifiers, and few-shot prototypical networks. Generalization patterns vary: supervised and few-shot models excel on seen artists and complex prompts, whereas style descriptors transfer better when the artist's style is pronounced; multi-artist prompts remain the most challenging. Our benchmark reveals substantial headroom and provides a public testbed to advance the responsible moderation of text-to-image models. We release the dataset and benchmark to foster further research: https://graceduansu.github.io/IdentifyingPromptedArtists/

Problem

• Identifying artist names from text-to-image generated pictures
• Benchmarking recognition of prompted artists across diverse settings
• Evaluating methods for responsible moderation of text-to-image models

Innovation

• Benchmark for prompted-artist recognition from images
• Evaluates multiple methods including supervised classifiers
• Dataset spans diverse generalization settings