Evaluating Intellectual Property Guardrails of Generative Image Models: A Technical Report

2026-06-30

AI Summary

This study addresses the risk that generative image models may inadvertently reproduce intellectual property (IP)-protected content—such as fictional characters, celebrity likenesses, and commercial logos—despite existing safeguards of uncertain efficacy. To systematically evaluate this issue, the authors introduce the first comprehensive benchmark and automated assessment framework targeting multiple IP categories, combining prompt injection techniques with image recognition to conduct large-scale testing across 14 state-of-the-art text-to-image models. Results reveal that all proprietary models exhibit some degree of IP refusal behavior, though protection strength varies significantly; notably, commercial logos remain the most easily generated, with all models still capable of producing recognizable protected content as of March 2026. This work provides a reproducible, quantitative methodology and empirical foundation for assessing IP safety in generative models.

Abstract

Generative image models are capable of producing images that bear a strong resemblance to, or replicate, recognizable intellectual property (IP). In this technical report, we present a benchmark and automated evaluation pipeline to test for evidence of IP guardrails in generative image models along with the propensity for these models to generate images with recognizable IP. The IP categories we tested include fictional characters, celebrity likeness, and commercial logos and do not encompass the full range of IP which may be implicated by image generation models. We evaluated fourteen widely used text-to-image models, including three self-hosted open weights models and eleven private models. While all of the private models were observed to refuse generations at some level due to IP guardrails, the frequency of generation refusals varied substantially among models. The refusal rates also varied considerably across the different IP categories tested. Commercial logos were refused least frequently and were successfully generated at the highest rate, on average. Though the rate varies, all models tested readily generated images containing recognizable IP as of March 2026.

Problem

• intellectual property
• generative image models
• IP guardrails
• text-to-image generation
• copyright infringement

Innovation

• IP guardrails
• generative image models
• automated evaluation pipeline
• benchmarking
• intellectual property