A model shoots a routine catalog job against a plain white background. Two years later, her face is appearing in ads for scenes she never shot, in poses she never held, generated by AI. Whether the retailer had the right to do that comes down to a few words of boilerplate that nearly every working model has signed without a second thought. That is Pujols v. Rainbow Shops in one paragraph, and it is the case every model, designer, and small brand working anywhere near AI should be following. This is the full story so far: the allegations, Rainbow's contract defense, the first ruling, and what it all means for the release sitting in your drawer.

The story so far

Francheska Pujols, a New York based model, shot a freelance catalog job for Rainbow USA, the budget retailer behind Rainbow Shops, in September 2024. Plain white background, standard poses. She alleges that Rainbow later used generative AI to place her likeness into images she never made: seated with her legs apart on a barstool holding a cocktail, resting her head in another model's lap, standing barefoot on a couch, and at least one image exposing her undergarments. She calls some of the pictures crude and says they damage her standing as a high-end model. She sued in New York Supreme Court on May 22, withdrew the case a week later to pursue a private settlement, and refiled on June 15 when those talks failed. The refiled complaint, docketed as Pujols v. Rainbow Shops, Inc., No. 157543-2026, added a new piece of evidence: an Instagram post showing her in a pose her lawyer says she never approved.

This was never a one-off grievance. Business Insider reported that Rainbow's studio manager warned freelance models in June 2025 that the company had begun 'styling certain products, and generating avatars, with the assistance of AI,' and that fewer people would be needed going forward. Other models say AI look-alikes of them cropped up across Rainbow's site, social media, and newsletters around the time their bookings dried up. A second model, Elii Emeghebo, has separately accused a retailer of using AI to alter his features, including lightening his skin tone, a claim that layers discrimination questions on top of the licensing dispute. Rainbow denies any wrongdoing.

Rainbow's defense

Rainbow's July 8 memorandum is the most important document in the case, because it is the argument every brand will copy if it works. Rainbow does not claim AI changes anything. It claims the contract already settled the question. The 2024 Freelance Model Waiver/Release/Indemnification Agreement, the company says, grants it the right to use Pujols' images 'whether intact or in part, composite or distorted in character or form, cropped or altered, without restriction as to changes or transformations,' waives her right to inspect finished advertising, and releases Rainbow from liability for 'blurring, distortion, alteration, optical illusion, or use in composite form.' On that footing, Rainbow has moved to dismiss her claims under New York Civil Rights Law Sections 50 and 51, the Lanham Act, and related state theories, relying on decades of New York precedent enforcing model releases according to their plain language.

Pujols reads the same words very differently. In an affidavit, she states that her contract only covered images taken during photo shoots and 'does not in any way authorize the creation of entirely new images, scenes, poses, or compositions that did not exist in the original content.' She thought the deal was 'pretty normal' when she signed it.

Round one: the injunction

On July 16, Justice James G. Clynes of the New York Supreme Court denied Pujols' motion for a preliminary injunction. The disputed images stay up while the case proceeds. That ruling decides nothing about who is ultimately right: emergency relief requires a showing of immediate, irreparable harm that is very hard to make before discovery, and the denial is procedural, not a judgment on the merits. But it matters in practice. Whatever harm Pujols is suffering keeps accruing while litigation moves at litigation speed, and the status quo now belongs to Rainbow.

Why this case matters

This is not a training-data case, and that distinction matters more than it might seem. As one firm tracking the dispute put it, the case is not about training AI models or scraping copyrighted materials; it is about the downstream commercial use of allegedly unauthorized synthetic depictions of a person whose likeness had already been licensed. Nobody disputes that Rainbow could photograph Pujols and run those photographs. The fight is over what happens after the shoot, once a brand feeds a model's licensed images into a generative tool and produces something new. Words like 'minor edits,' 'alter,' and 'composite' were drafted for cropping, color correction, and retouching. Generative AI can take a face from a white background and put it anywhere, in any posture, doing anything. So the question the court has been handed is the one the whole industry needs answered: when does an authorized modification of a licensed image become an entirely new depiction requiring separate permission? If AI composites are just another form of editing already licensed under old language, the practical protection for talent shrinks to whatever specific words happen to appear in releases signed years before these tools existed. If a materially new depiction, showing a person in acts, settings, or states of undress they never performed, exceeds any reasonable reading of 'edit,' the burden shifts back to brands to obtain fresh, specific consent.

The new laws in the background

New York changed the ground under this case while it was pending. The Fashion Workers Act, championed by the Model Alliance, took effect June 19, 2025, and requires clear written consent, separate from the underlying release, before a company creates or uses a model's digital replica, with the consent spelling out scope, purpose, compensation, and duration; routine retouching is carved out. The state's AI Transparency in Advertising and Synthetic Performer Disclosure Law followed on June 9, 2026, requiring a conspicuous disclosure whenever a commercial ad features an AI-generated synthetic performer, with civil penalties for skipping the label. Neither statute reaches back to a release signed in 2024, which is exactly why this lawsuit is being watched so closely: it tests the gap years, when the only protection anyone had was whatever the parties actually wrote down.

The parallel case: Tranchin v. EBY

Pujols is not litigating this question alone. In Tranchin v. EBY, Inc., filed June 9, 2026 in the Northern District of California, influencer Molly Tranchin, known as FashionVeggie, alleges the intimate-apparel brand EBY used AI to turn footage she supplied into an explicit video and posted it to Instagram without her review or consent. Her complaint brings claims under California's nonconsensual deepfake statute alongside privacy, defamation, contract, tortious interference, and copyright theories. Two working models, two coasts, one unglamorous question: what rights did the brand actually purchase? If Rainbow's contract defense survives its motion to dismiss, expect it to be cited in the EBY case within days.

What to watch

First, the motion to dismiss. A grant would hand brands a template defense for AI-altered marketing built on existing releases, at least in New York, and would tell every company sitting on old catalog imagery that decade-old boilerplate might double as an AI license. A denial opens discovery into which AI tools were used and who directed the outputs, a record that will shape every future case of this kind. Second, whether more models come forward. If other Rainbow models join or file their own claims, this stops being an isolated dispute and starts looking like an industry practice under scrutiny. Third, your own paperwork, because the practical lesson is available now, regardless of how the case ends. If your release grants rights to 'alter,' 'transform,' or 'composite' without naming AI, assume a court might read it more broadly than you intended. Name AI specifically. State whether synthetic, composited, or AI-generated depictions are permitted at all, and if so on what terms, with what approval, and for how long. Demand the separate digital-replica consent New York law now requires. Silence in a contract is never neutral. Right now, in a Manhattan courtroom, it may be doing exactly what Rainbow says it always could.

Sources