Managing a Model's Digital Likeness After the Campaign: Tracking, Renewals, Takedowns
Managing a model's digital likeness after a campaign means keeping the licence alive as a working record, not filing it away. Four things need an owner. First, a usage register that lists every asset made with the likeness, where it runs and which licence term covers it. Second, a calendar of expiry dates, with a decision on each: renew, let lapse or take down. Third, a takedown routine for the day a licence ends, consent is withdrawn or an image appears where it was never agreed. Fourth, a deletion record that shows the likeness files and training material were removed when the contract said so. Most disputes start after launch, when nobody can say which images are still live. The legal duties are set by the contract and by the law of each market, so the register is the tool that lets a brand answer a model's agency, a regulator or its own counsel quickly. This article is not legal advice.
Consent and licence terms before production are covered in Model Likeness Rights in AI Campaigns, and how a twin is made and paid in Digital Twins of Real Models. This article starts where those end: the campaign has launched.
Why the work starts after launch
A shoot produces a fixed set of photographs. A licensed likeness can produce new images for as long as the licence runs, and each image can be cropped, resized, translated into new formats and reposted by retailers, regional teams and agencies. After a few months the answer to "where is this face being used?" is spread across drives, ad platforms and partner portals.
The licence, meanwhile, has edges. It names a purpose, media, territories and a period. If you cannot tell which asset sits inside which edge, you cannot tell when you have stepped outside it. Platforms and partners will not warn you. The model's agency may ask, and a regulator or a journalist may too, and "we think it ended in March" is a weak answer.
The usage register
A usage register is a list with one row per asset, or per asset family where variants share the same terms. It does not need special software; a shared spreadsheet that one named person keeps current is enough. Each row should record:
Asset and version. A file name or ID that matches what the studio delivered, including approved variants.
Licence reference. Which booking or licence the asset rests on, and the page or clause that sets its scope.
Where it runs. Channel, market and first publication date, including partner and retailer use.
End date. The last day the asset may run under that licence, and the date by which it must be pulled.
Approval record. Who approved the image, when, and whether the model or agency saw it, if the licence asks for that.
Labelling. Whether the AI label is in place wherever the asset runs. The rules for that are in EU AI Act Labelling Rules for AI Ad Imagery.
Status. Live, scheduled for takedown, taken down, or extended.
The studio and the creative agency should feed this list, because they know which variants exist. The brand should own it, because the brand is the party that publishes. The review and approval workflow is the natural place to start the register, since every approved asset already passes through it.
Renewals and expiry
Put every end date in a calendar, with a reminder far enough ahead to decide. What "far enough" means depends on the contract: if an extension needs the model's confirmation, the reminder has to leave room for the agency's response time plus your own approval round.
At each date there are three decisions.
1. Renew or extend. An extension into a new season, market or medium is a new request, not an automatic one, as set out in Licensed Digital Twins vs Synthetic AI Models. Ask early, state the new scope and keep the answer with the register.
2. Let it lapse. Pull the assets from every channel, including partners, by the end date. Agree in the contract whether a short sell-off or run-out period exists, and write down the date it applies.
3. Replace. If the campaign continues without the likeness, the brief to the studio changes. That belongs in the next production round, not in a scramble on the last day.
Two traps recur. The first is the licence for the likeness ending before the licence for the images: a clean contract states both dates separately. The second is paid media. Boosted posts, retargeting and partner placements run on their own schedules and are easy to miss when the organic post has long gone.
Takedowns: who does what, how fast
A takedown request can come from three places: your own calendar, the model or their agency, and the outside world, where someone has copied or altered an image.
Expiry and withdrawal. Agree in advance who sends the instruction to channels and partners, in what form, and how quickly each one must act. A one-page takedown checklist with channel owners, partner contacts and a confirmation step (screenshot or link check) turns a vague promise into a routine. A model's consent can end earlier than the licence. Under the GDPR, a person can withdraw consent at any time, and withdrawal must be as easy as giving it (Art. 7(3)); the earlier processing stays lawful. In a commercial licence, the contract decides what happens to images already in circulation, and a model agreement written for AI use should say it plainly.
Misuse by third parties. A fake ad, an altered image or a copy on a site you do not control is a different problem. Steps that usually apply:
Save the evidence first: URL, date, screenshots and the original file.
Tell the model's agency. The model should hear it from you, not from a friend.
Use the platform's reporting route. In the EU, hosting services must offer a mechanism for notices of illegal content (Digital Services Act, Art. 16). Other markets have their own routes, and platform rules often move faster than court action.
Ask counsel about claims in the market concerned.
Deletion: what must be removed and how to show it
Once a licence ends or consent is withdrawn, three kinds of material need a decision: the capture material and training set, the likeness files themselves, and the delivered assets. The first two are normally deleted at the end; the third follows its own usage period. Digital Twins of Real Models explains that split. Here, the question is proof.
Under the GDPR, a person has the right to erasure without undue delay where, among other grounds, they withdraw consent and no other legal ground for the processing remains (Art. 17). Exceptions exist, among them the establishment, exercise or defence of legal claims. What that means for a specific likeness workflow is a question for a data protection lawyer.
Whatever the legal basis, a deletion record is worth having. It should state:
what was deleted: training images, trained model, test outputs, working files
where it was held, including backups and subcontractors
who deleted it, on which date, and who confirmed it
what was kept, such as delivered assets under licence, invoices and the consent record, and why
Ask for it in writing at the end of the contract. If the studio cannot name every place the material sat, that is worth knowing before the licence ends, not after. Sharing Assets With an AI Studio covers the access and storage terms that make this possible.
Where the law is moving
Rules on digital likenesses are changing in several markets, and the position needs checking against the date and the country of each campaign. As of 20 November 2026:
United States, federal. The NO FAKES Act would create liability for producing or distributing an unauthorised digital replica of a person's voice or likeness. It has been introduced in several versions; the 2025 bills are S. 1367 and H.R. 2794, and a 2026 Senate version, S. 4591, was reported to have cleared the Judiciary Committee in June. We could not confirm that any version has become law. Check the bill page before relying on it.
United States, takedown of intimate deepfakes. The TAKE IT DOWN Act requires covered platforms to remove non-consensual intimate imagery, including AI-generated imagery, within 48 hours of a valid request. The Federal Trade Commission began enforcing it in May 2026. It covers a narrow category and does not make every misuse of a likeness removable on request.
New York and California. New York's Fashion Workers Act requires written consent for digital replicas of models, covering scope, purpose, rate of pay and duration. California's Labor Code § 927 can make a vague digital replica clause unenforceable. Both turn the licence terms, and the record of staying inside them, into evidence.
EU. The GDPR rules above apply wherever a likeness of an identifiable person is processed. Labelling duties for AI-generated imagery come from the AI Act and are covered separately.
This is a snapshot, not a forecast. Laws like these change often.
A one-page routine to adopt
1. Name one owner at the brand for the register and one contact at the studio and the model's agency.
2. Fill the register before launch, not after.
3. Put every end date in a shared calendar with a reminder long enough for a renewal round.
4. Keep a takedown checklist with channel owners and partner contacts.
5. Ask for a written deletion record at the end of the contract.
6. Review the register at every campaign close and every quarter while assets run.
Licensing terms and category protection for licensed twins are set out on the page about licensing and rights. If you are planning a campaign with a real model's likeness, explore licensed digital twins.
Planning a campaign? Request an AI campaign and we get back to you.
Professional model with an agency? Apply as a digital twin.
Are you a professional model and want to apply as a digital twin? Apply here.
Frequently Asked Questions
What is a usage register for a model's digital likeness?
It is a list of every asset made with the likeness, with the licence it rests on, where it runs, its end date, its approval record and its status. It lets a brand show quickly which images are live and which are covered. One named person should keep it current.
What happens to AI campaign images when a licence for a likeness expires?
The images should stop running when the licence for them ends, in every channel including partners and paid media. The contract decides whether a run-out period exists and what happens to material the brand already holds. Write both dates, for the likeness and for the images, into the agreement.
Can a model withdraw consent after a campaign has launched?
Under the GDPR a person can withdraw consent at any time, and earlier processing stays lawful. In a commercial licence the contract sets what happens to images already in circulation. Ask for a clause that names the takedown steps and deadlines before you sign.
Does the right to erasure force a studio to delete a model's likeness files?
Where a person withdraws consent and no other legal ground for processing remains, the GDPR gives a right to erasure without undue delay, with listed exceptions such as legal claims. Whether a given workflow falls under that depends on its design, so get a data protection assessment.
How should a brand respond when a deepfake of a campaign model appears?
Save the evidence, tell the model's agency, report the content through the platform's route and ask counsel about claims in the relevant market. In the EU, hosting services must provide a notice mechanism. In the US, the TAKE IT DOWN Act covers intimate imagery only.
Is there a US federal law on digital replicas?
As far as we could confirm on 20 November 2026, the NO FAKES Act has been introduced in several versions but is not law. State rules, such as New York's Fashion Workers Act, and contracts apply meanwhile. Check the bill status before relying on it.
Who should own digital likeness management at a brand?
The brand, because it publishes the images. The studio and the creative agency supply the asset information, and the model's agency receives the notices. One named owner on each side keeps the register and the calendar from falling between teams.
About The Bridges AI
The Bridges AI is a hybrid AI creative agency: a creative studio and production partner for fashion, beauty, luxury and hospitality brands and for the agencies that work with them. It combines booked real talent, real products and human creative direction with AI production, and offers licensed digital twins of real models, only of models under contract. The Bridges AI works in strategic partnership with Most Wanted Models®, the German model agency founded in 1997, which brings nearly 30 years of experience in bookings, buyouts and usage rights.
Written by the editorial team of The Bridges AI. Last updated November 2026.




Comments