How to set up consent, payment and disclosure for AI digital twins

Fashion retailers deploying AI digital twins of models must navigate three pillars: securing informed consent through contracts that address generative AI, structuring fair payment for ongoing digital use, and meeting EU transparency requirements that mandate watermarking from August 2026. Brands that fail to address these elements risk regulatory penalties, talent disputes and consumer trust issues as synthetic imagery becomes standard across e-commerce and marketing.

What does Article 50 of the EU AI Act require for digital twins?

The EU AI Act, which entered into force in August 2024, imposes obligations on deployers of AI systems that generate synthetic content depicting people. Article 50 requires that artificially generated or manipulated image, audio or video content be marked in a machine-readable format and disclosed as artificially generated when published. The transparency requirements apply from 2 August 2026, giving fashion brands a compliance deadline for all AI-generated imagery featuring recognisable individuals.

Watermarking must be visible or otherwise detectable to inform viewers that the content is synthetic. This applies to digital twins created from model likenesses and used in catalogues, social media, advertising and virtual try-on tools. Brands deploying AI-generated models must implement technical solutions that embed markers without degrading image quality or disrupting creative direction. Failure to comply can result in fines of up to 1.5 per cent of global annual turnover or 7.5 million euros, whichever is higher.

How should model contracts address digital twin rights?

Traditional modelling contracts grant usage rights for specific photoshoots, campaigns and durations. Digital twin agreements must go further, explicitly covering the creation, storage and reuse of AI-generated likenesses. Key contract clauses include scope of use (which product categories, marketing channels and geographies), duration (one campaign, seasonal, perpetual), exclusivity (whether the twin can be licensed to other brands), and revision rights (whether the brand can alter the digital likeness without further consent).

Models should retain the right to approve initial digital twin creation and review how their likeness will be used. Contracts should specify whether consent covers future campaigns or requires renewed approval for each deployment. Legal experts recommend separating digital twin rights from physical shoot rights, with distinct payment terms for each. Models may also negotiate limits on use cases, excluding contexts such as controversial products or markets where they prefer not to appear.

What payment structures work for digital twin licensing?

Payment models for digital twins vary based on scope, duration and bargaining power. Common structures include one-off buyouts, annual licensing fees, per-campaign payments and revenue-share arrangements. One-off buyouts grant perpetual rights in exchange for a lump sum, typically higher than traditional shoot fees to reflect ongoing value. Annual licensing involves recurring payments for continued use, with fees adjusted for exclusivity and territory.

Per-campaign payments treat each deployment as a separate transaction, similar to traditional usage rights. Revenue-share models tie compensation to measurable outcomes such as impressions, conversions or sales attributed to campaigns featuring the digital twin. Hybrid structures combine upfront fees with performance bonuses. Payment levels depend on the model's profile, market rates and whether the brand retains exclusive rights or shares the twin with other licensees.

• One-off buyout: single payment for perpetual, exclusive use across all channels and geographies.
• Annual licence: recurring fee for defined scope, renegotiated or renewed each year.
• Per-campaign fee: separate payment for each campaign, product line or seasonal collection.
• Revenue-share: percentage of sales or cost-per-impression tied to campaigns using the digital twin.
• Hybrid model: upfront fee plus performance bonuses based on usage metrics.

Which watermarking methods meet compliance requirements?

Technical watermarking for AI-generated fashion imagery falls into two categories: visible labels and invisible metadata. Visible watermarking includes text overlays, logo badges or corner icons that signal synthetic content directly to viewers. These must be legible and persistent across platforms, including social media and third-party resellers. Invisible watermarking embeds machine-readable information in image files using steganography or metadata standards such as IPTC or C2PA (Coalition for Content Provenance and Authenticity).

The C2PA standard, supported by Adobe, Microsoft and the BBC, embeds cryptographic hashes and provenance data that track content origin and modifications. Fashion brands can use C2PA-compliant tools to attach certificates confirming AI generation, ensuring compliance even if images are cropped or compressed. Visible and invisible methods are often combined: a discreet icon satisfies consumer-facing transparency while metadata enables platform-level detection and regulatory audits.

How can brands implement consent and disclosure workflows?

Operationalising consent and disclosure requires coordination across legal, creative, IT and marketing teams. Brands should establish a contract review process that flags digital twin clauses, ensuring models receive clear explanations of AI use cases before signing. Legal teams should maintain a rights database tracking which models have consented to digital twin creation, the scope of permitted use and expiry dates for time-limited agreements.

Creative teams must integrate watermarking into asset production pipelines, applying labels and metadata before images enter digital asset management systems. Marketing teams should document where digital twins appear, enabling compliance audits and ensuring disclosure language accompanies synthetic content on websites, apps and print materials. Regular training on EU AI Act obligations and emerging case law helps prevent inadvertent violations as regulations evolve.

What are the risks of non-compliance or ambiguous contracts?

Brands that deploy digital twins without clear consent face legal claims from models for unauthorised use of likeness, potentially resulting in injunctions, damages and reputational harm. Ambiguous contracts that fail to specify AI use can lead to disputes over payment, scope and exclusivity, especially if models later object to how their digital twin is deployed. Non-compliance with Article 50 transparency requirements exposes brands to regulatory fines and enforcement actions from national authorities under the EU AI Act.

Consumer trust is also at stake. Undisclosed use of synthetic models can trigger backlash if audiences perceive deception, particularly around body image and diversity representation. Transparent disclosure, coupled with fair compensation and robust consent, mitigates these risks and positions brands as responsible adopters of generative AI technology in fashion marketing and e-commerce.