licensefoundry

License Exposure Review

Your AI exposure isn't the license you signed. It's what you can't prove it covers.

You signed it. The money moved. The terms were probably reasonable. But if a counterparty disputed the scope — or a regulator asked you to evidence it — what exactly would you produce, and would it hold?

A ten-point review of your existing arrangements: grant granularity, counterparty binding, asset identification, revocation mechanics, upstream chain of title, and whether your Article 4 reservation posture matches what your systems actually publish. Written findings, rated, with remediation in priority order.

Start with an Exposure Scan · €2,500 · five working days

A structural and evidentiary review, not legal advice.

Who this is for

Organisations that have already signed content licensing arrangements. The money is spent, the deal exists, and what you hold is a PDF and a spreadsheet. You do not need convincing that the problem is real — you need someone to tell you how bad yours is.

  • Reproduction Rights Organisations and collecting societies licensing on behalf of others, where chain of title is existential.
  • News publishers who have signed lab deals, whose boards are now asking questions the deal team cannot answer.
  • Broadcasters and AV archives with existing rights departments and heavy regulatory exposure.
  • Stock and specialist content libraries — photo, music, academic.
  • The buy side: European companies fine-tuning or deploying models on acquired content.

The ten-point framework

The same framework runs every engagement. That is deliberate: it is what keeps this a repeatable review rather than bespoke consultancy, and it is why the findings are comparable across agreements and across years.

  1. Grant granularity

    Does the agreement enumerate permitted uses at the resolution AI requires — training, fine-tuning, retrieval-augmented generation, embedding, output derivation, inference-time use — or does it say “use” and “reproduce”?

  2. Counterparty binding

    Which legal entity holds the grant? Does it extend to affiliates, successors, subprocessors, or anyone receiving the model weights? If the counterparty open-weights the model, does the grant travel with it?

  3. Asset identification

    How is the licensed corpus identified? Content hashes and a manifest, or “the Publisher's catalogue”? Could you demonstrate in 2030 which specific assets were in scope on the day of grant?

  4. Temporal proof

    Can you evidence what was granted on what date, against which version of the corpus and which version of the agreement?

  5. Revocation and termination mechanics

    On termination, what happens to a model already trained? Is there a mechanism, or only a clause? Most agreements are silent here, and the silence is catastrophic.

  6. Upstream chain of title

    Did the grantor actually hold what it granted? Freelance contributions, work-for-hire gaps, archival material, third-party imagery, syndicated content.

  7. Machine-readable reservation posture

    Does the Article 4 reservation (Directive 2019/790) match what your systems actually publish — robots.txt, TDMRep, RSL, terms of service? The Dutch HowardsHome ruling (ECLI:NL:RBAMS:2024:6563, under appeal) held that naming a few bots in robots.txt was not a valid reservation; it is a TDM ruling whose machine-readability holding also governs AI-training opt-outs.

  8. AI Act Article 53 alignment

    Can the counterparty produce the training-data summary this agreement implies? Does your own record corroborate or contradict what they would file?

  9. Audit and inspection rights

    Are there any? Can they be exercised in practice, or only in theory?

  10. Evidentiary survivability

    The synthesis. If a hostile reader challenged this in 2030, what precisely would you produce, and what would they do with it?

Every point is rated. The rating table is the first page you read, and it is the page you forward internally.

SOUND WEAK ABSENT

Scope and price

Engagement Price Duration Scope
Exposure Scan €2,500 5 working days One agreement, one counterparty. Two-page written verdict. Credited in full against a Review commissioned within 60 days.
License Exposure Review €15,000 3 weeks Up to 8 agreements. Full report against the ten-point framework, plus a 90-minute findings session.
Portfolio Review €30,000 6 weeks Up to 25 agreements, upstream chain-of-title sampling, remediation roadmap, board-ready summary.

Available in Europe only — not currently offered in the USA. Reviews are priced in euro. The issuance platform is priced separately in US dollars — different products, different buyers, and the platform carries no such restriction. See platform pricing →

Payment terms

  • Exposure Scan: 100% on commission, before work starts.
  • Review and Portfolio Review: 50% on commission, 50% on delivery. Payment within 14 days.
What this is, and what it is not. The Review is a structural and evidentiary assessment of the documents and systems you already have: what your arrangements say, what your systems publish, and what you could produce if asked. It is not legal advice, it is not an opinion on the validity or enforceability of any agreement, and it does not determine whether you or any counterparty holds the underlying rights. Where a finding turns on a question of law, the report says so and identifies it as a question for your counsel.

Start with an Exposure Scan.

One agreement, one counterparty, five working days, a written verdict on what your record would actually support. €2,500, credited in full if you commission the fuller Review within 60 days.

info@licensefoundry.com Check what your content declares →

A structural and evidentiary review, not legal advice. LicenseFoundry is a Netherlands-registered company; engagements are contracted from the Dutch entity.