The AI Lawsuit That Every Jewellery Brand Using Model Photography Should Read
- 13 minutes ago
- 8 min read
In May 2026, New York model Francheska Pujols filed a lawsuit against budget clothing retailer Rainbow Shops in the New York Supreme Court. The claim was straightforward: Rainbow had photographed her wearing their clothes against a plain white background, then used AI to generate entirely new images, including one with her legs spread over a barstool and another of her resting her head on another model's lap while holding a cocktail.
The original contract allowed minor photo edits, not new scenes, new poses or a different version of her body placed into contexts she had never agreed to. When the contract expired, Rainbow reportedly continued using her likeness. She sent a cease and desist in March 2026. It was ignored.
Rainbow's public response:
"We used our images properly and there's no violation of her rights."
Both parties are now seeking a private settlement.
What This Actually Is: AI Likeness Generation Without Consent
AI likeness generation is the use of generative AI tools to create new images of a real person — using existing photographs as a reference — without that person posing for the new image. The person does not know what the image will look like. They do not approve it. They may not even know it exists.
This is distinct from:
Colour correction — adjusting tones in an existing photograph
Background replacement — placing a product against a different backdrop
Retouching — removing a blemish or adjusting lighting in a shot that was taken
Likeness generation creates a new image. The model's face, body proportions, and identifying features are reproduced in a scene that never happened.
The distinction matters legally, commercially, and reputationally.
Why This Is a Problem Now — and Why the Fashion Industry Is Not a Safe Distance Away
Let's be direct about what happened in the Rainbow Shops case. A brand photographed a model legitimately. The model was paid, the shoot was contracted, the images were used — so far, so normal.
But then the brand fed those photographs into an AI system and generated new content: new scenarios, new compositions, new poses. Content that the model described as damaging to her professional image as a high-end talent.
The brand's reasoning, presumably, was cost. Generating a new image via AI costs a fraction of rebooking a model, rescheduling a shoot, and editing the results.
Fashion law expert Anthony Lupo, speaking to the New York Post, shared:
"The less established models are going to have real problems making a living just doing modeling."
He estimated that 85% of commercial modelling — the day-to-day "I need someone to wear this" category — is already at risk of being replaced by AI-generated images. The supermodel tier survives. Everyone else faces a structural collapse in demand.
This is not a future concern. H&M began rolling out AI-generated model clones in 2025. J.Crew has been producing AI-generated fashion images that are, according to observers, difficult to distinguish from real photographs.
Strategic business attorney Joshua R. Bressler described the current legal landscape as "coming fast and furious" — a system actively sorting itself out in real time.
The jewellery industry sits inside this exact same dynamic. Jewellery brands use model photography. And increasingly, they are asking whether they can use AI to reduce the cost of doing all of that.
The answer is: yes, conditionally. But the conditions are tightening — fast.
The Regulatory Environment: What the US and EU Are Actually Requiring
This is where it becomes operationally relevant for jewellery brands, particularly those selling into the EU or the US market.
United States: State-Level Laws with Real Teeth
There is no single federal AI likeness law in the US — yet. The proposed NO FAKES Act would create a nationwide right against unauthorised digital replicas of a person's likeness, voice, or performance, but it has not been enacted.
What does exist, and is in force:
New York Fashion Workers Act (effective 19 June 2025): Requires clear written approval from a model before creating or using their digital replica. The consent must specify scope, purpose, rate of pay, and length of time. Routine retouching is excluded — but AI generation of new scenarios is not.
New York Digital Replica Law (effective 1 January 2025): Any digital replica agreement is void if it allows a replica to be used instead of work the person would have done in person, fails to give a specific description of the intended use, and was negotiated without legal counsel available to the individual.
New York AI Transparency in Advertising and Synthetic Performer Disclosure Law (effective 9 June 2026): Requires disclosure in commercial advertising that uses AI-generated digital performers.
Tennessee's ELVIS Act (effective 1 July 2024): Extends publicity rights explicitly to AI-generated likenesses and voices, with civil — and in some cases criminal — penalties for unauthorised commercial use.
Colorado AI Act: Risk and impact assessments required for high-risk AI systems, enforcement from February 2026.
The pattern is consistent: written consent, specific scope, documented compensation, and disclosure. Vague language in a contract no longer provides cover.
European Union: The AI Act Is Now In Force
The EU's Artificial Intelligence Act entered into force on 1 August 2024. Its provisions are becoming applicable in stages:
Prohibited AI practices: applicable since 2 February 2025
General-purpose AI model rules: applicable since 2 August 2025
The majority of AI Act provisions: applicable from 2 August 2026 — which is now weeks away
In September 2024, the European Commission published guidelines on prohibited AI practices. The transparency rules of the AI Act come into full effect in August 2026. A political agreement on the 'AI omnibus' simplification package was reached in May 2026, which extended the transition period for certain high-risk embedded systems — but the core obligations for commercial AI use remain on schedule.
The EU also requires that AI-generated content be clearly labelled when it can mislead consumers. The General-Purpose AI Code of Practice, finalised in July 2025, covers transparency, copyright, and safety.
Under GDPR, which is already fully in force, using a recognisable person's likeness in generated content without a legal basis — such as explicit consent — is a data protection violation as well as a commercial one.
Jurisdiction | Key Law | Status | Applies To |
New York (US) | In force since June 2025 | Brands using model photography | |
New York (US) | In force since Jan 2025 | Contracts with models | |
New York (US) | In force since June 2026 | Commercial advertising | |
Tennessee (US) | In force since July 2024 | AI likeness and voice use | |
EU | From August 2026 | All AI systems deployed commercially | |
EU | In force since Feb 2025 | High-risk and manipulative AI |
What This Means for Jewellery Brands Using AI Photography
Jewellery brands — particularly independent and mid-scale brands selling into the EU or US — are often using AI image generation in one of three ways:
Replacing background scenes in product-only photography (low risk, generally permissible)
Generating AI models to wear jewellery instead of hiring human models (medium risk, depends on origin of training data)
Taking photographs of real, named or contracted people and using AI to generate new scenarios with their likeness (high risk — this is the Rainbow Shops situation)
The third category is the one with immediate legal and reputational exposure.
If your brand has a signed model, a contract that permitted photography and minor editing, and you then use that photography as a prompt to generate entirely new images — you are in materially the same position as Rainbow Shops. Your contract likely does not cover this. The laws in New York and the EU do not permit it without specific, documented consent.
The second category carries a different risk, but it is more specific than it first appears. The legal question is not whether the AI tool was trained on licensed data — that is a dispute between the tool provider and the courts, not between you and a claimant. The actual legal test is identifiability: if the generated face is genuinely new and no real person could recognise themselves in it, there is no likeness to protect and no claimant under existing law. If, however, the output closely resembles a specific real individual — a model whose reference photo was in your moodboard, for instance — right-of-publicity protections apply. The distinction between generating a new face and replicating a real one is the line the law is currently drawing.
The first category — background replacement, product placement on AI-generated abstract or generic environments — is the most defensible, provided you are not generating a recognisable human likeness.
What Jewellery Brands Should Do With This Information
This is not about halting AI photography use. It is about using it with correct documentation and within the current legal framework. Here are the concrete steps:
1. Audit Your Existing Contracts with Models
Every contract signed before June 2025 predates the New York Fashion Workers Act. If you have photography contracts from that period and you are using — or plan to use — those images as inputs for AI generation, those contracts almost certainly do not include the required consent clauses. Using those images for AI generation is legally ambiguous at best. Get updated agreements.
2. Update Model Release Language Now
Any new photography contract should include explicit language covering:
Whether the brand may use the images as inputs for AI generation
What AI-generated outputs are permissible (background replacement vs. likeness generation)
Duration and scope of AI use rights
Compensation for AI use is separate from photography session fees
A standard photography release does not cover this. A clause added in five sentences does.
3. Distinguish Between Product Photography and Likeness Generation
Product photography — a ring on a hand, a necklace against a background — is categorically different from generating a new image of a named model in a new context. If you are doing AI-assisted product photography without generating a recognisable human likeness, your risk profile is much lower. Keep a clear operational distinction between the two.
4. Document the AI Tools You Are Using and Their Training Data Policies
If you use an AI image generator for commercial outputs, check whether that tool has published information about its training data, licensing, and compliance with EU AI Act transparency requirements. Tools that have signed the EU GPAI Code of Practice have a presumption of conformity with the Act — a meaningful practical distinction when selling into the EU.
5. Label AI-Generated Content in EU Markets
From August 2026, the EU AI Act's transparency provisions require that AI-generated content be clearly identified. For jewellery brands operating in the EU, this means product images generated by AI should be labelled as such. The format and placement requirements are still being defined in implementation guidance, but the principle is established.
6. Work With a Creative Partner Who Understands Both AI and Legal Exposure
One of the genuine advantages of working with a creative agency that specialises in jewellery photography — rather than a generalist AI tool or a freelancer unfamiliar with commercial content law — is that the liability question is handled as part of the workflow, not discovered after the fact.
At Chocianaite, we produce both real jewellery photography and AI-generated jewellery imagery for brands. The distinction between what we generate and what requires human model involvement is built into how we scope each project. When human models are used, contracts are drafted to reflect current legal requirements. When AI-only imagery is produced, the method is documented and disclosed. Neither category requires a lawsuit to sort out.
The Bottomline
The Rainbow Shops case was not about whether AI photography is legitimate. It was about whether a brand can take a real person's likeness and generate new images of her without consent — and then continue doing so after the contract expired.
Courts and regulators in both the US and the EU are moving toward the same answer: no, not without documented, specific, compensated consent.
The jewellery industry is not exempt from this dynamic. Brands using model photography — AI-assisted or otherwise — are operating in a regulatory environment that is now actively enforced, not just proposed.
The brands that treat this as a workflow and compliance question now will have fewer difficult conversations later.
Comments