AI-Regenerated Copyright Infringement Continues to Rise
This month we have seen a significant rise in what can best be described as AI-regenerated copyright infringement. Rather than directly copying and re-uploading product photos from legitimate brands, illegitimate sellers are increasingly using AI tools to regenerate existing copyrighted images before using them in infringing listings.
The process is relatively simple. A seller takes an original product image owned by another company and runs it through an AI image generation tool. The AI then recreates the image with subtle modifications. It may smooth edges, alter lighting, place the product in a person’s hand or generate a slightly different background. At first glance, the image can appear “new”.
However, the core composition of the image often remains substantially the same. The product angle, positioning, framing and overall creative setup are frequently derived directly from the original copyrighted work. In many cases, these regenerated images are clearly designed to imitate the appearance and marketing style of the original brand while attempting to avoid traditional image matching systems and copyright enforcement tools.
This is creating a growing challenge for brands and marketplaces alike.
Historically, copyright enforcement has relied heavily on identifying direct image duplication. Exact or near-identical copies were relatively straightforward to detect and remove. AI regeneration changes that landscape considerably. Infringing content can now be altered just enough to bypass the marketplaces automated systems while still benefiting from the creative work and commercial value of the original image.
What has become increasingly clear is that copyright law and platform enforcement processes have not fully caught up with the pace of generative AI technology. Many marketplaces are still trying to determine where the line should be drawn between transformation and infringement.
Some platforms are making genuine efforts to adapt. Others continue to rely heavily on automated systems that struggle to assess image composition, creative similarity and contextual infringement. As a result, enforcement outcomes are becoming increasingly inconsistent across marketplaces.
From our perspective, the key issue remains straightforward. If an AI-generated image is clearly derived from a copyrighted product photo and reproduces the same creative composition for commercial gain, rights holders should still have a strong basis for enforcement.
As generative AI tools become more accessible, clearer copyright regulation and marketplace guidance will be essential. Without it, brands face a rapidly growing wave of harder-to-detect infringements that blur the line between copying and AI-assisted imitation.
What brands can do
Brands need to move beyond simple reverse image matching. AI-regenerated infringements are designed to avoid exact-match detection while still copying the original creative composition and commercial style.
Tools like IP Moat help brands detect manipulated listings by analysing products, image composition and listing patterns rather than relying solely on duplicate images. IP Moat also uses Computer Vision+ and intelligent product detection to identify infringements even when images have been altered.
An increasingly important part of enforcement is proving ownership at scale. The IP Moat Copyright Vault allows brands to securely store original product images, extract metadata, generate copyright certificates and create shareable proof of ownership for marketplace disputes.
As AI-generated infringement becomes harder to detect, brands that combine stronger copyright evidence with AI-powered monitoring and scalable enforcement tools will be in a far stronger position to protect their products online.
Marketplace Breakdown: Who’s Acting, Who’s Lagging
Compared to last month, overall marketplace response times remain broadly stable, with no major improvement across the industry. Amazon marketplaces continue to rank among the faster responders in most regions, while platforms such as Lazada, standalone websites and some social commerce channels continue to show slower enforcement turnaround times.
Notably, Website-hosted infringements improved slightly month-on-month, while Walmart Mexico and Wish showed slower average response times compared to the previous reporting period. TikTok Shop and social commerce platforms remain mixed, with enforcement speed still varying significantly by region and case complexity.
When grouped by marketplace category, traditional ads, gateways and secondary marketplaces continue to show the fastest average response times. In contrast, promotional platforms, print-on-demand services and standalone websites remain among the slowest areas for enforcement action.
Importantly, there has been no meaningful change in response times by IP type compared to last month, suggesting that marketplace enforcement speed is still being driven more by platform processes and internal moderation systems than by the specific type of intellectual property being reported.
