State of AI Identity Licensing 2026 methodology

A reproducible account of what the report reviewed, how each source was coded and what the resulting counts can and cannot support.

Version 1.1 · September 5, 2026. This method supports a bounded primary-source review. It is not a systematic review of every jurisdiction, case, contract or platform rule.

Research question

Which primary sources most directly describe the controls a buyer, talent representative or operator encounters when a real person’s face or voice is created or used as an AI digital replica?

Inclusion rules

What was excluded

Coding rule

Each source was coded against eight controls. A true value means the source expressly addresses the topic. It does not mean the source imposes a binding requirement, applies worldwide or is equivalent in authority to another source.

ControlCoding definitionSources
PermissionThe source addresses authorization, consent, licensing or a right to control use of a person’s voice or visual likeness.9 of 14
Specific scopeThe source addresses intended-use specificity, contract form, project limits or a requirement for additional approval.6 of 14
CompensationThe source addresses payment, wages, residuals, benefit contributions or disclosure of a paid relationship.4 of 14
Asset controlThe source addresses access, security, storage, transfer or destruction of a replica or source asset.2 of 14
DisclosureThe source addresses telling an audience that content is synthetic, manipulated, sponsored or otherwise materially connected.3 of 14
ProvenanceThe source addresses machine-readable origin, history, marking, binding, watermarking, detection or signed content assertions.3 of 14
LifecycleThe source addresses duration, expiry, termination, additional use, credential revocation or treatment after a permission period.7 of 14
RemedyThe source addresses unenforceability, liability, removal, a cause of action or a policy proposal intended to close a legal gap.7 of 14

Authority is not flattened

The dataset keeps enacted law, proposed legislation, government reports and guidance, collective agreements and technical standards in separate fields. A proposal is not reported as law. A collective agreement is not reported as universally applicable. A technical standard is not reported as personal permission.

Primary-source inventory

SourceAuthority classStatus
Copyright and Artificial Intelligence, Part 1: Digital Replicas
U.S. Copyright Office
Government report or guidancePolicy report and recommendations, not enacted law
AB 2602: Contracts against public policy, digital replicas
California Legislature
Enacted law or regulationEnacted, applicable to specified new performances fixed on or after January 1, 2025
AB 1836: Use of likeness, digital replica
California Legislature
Enacted law or regulationEnacted
ELVIS Act, Public Chapter 588
Tennessee General Assembly
Enacted law or regulationEnacted, effective July 1, 2024
S.1367: NO FAKES Act of 2025
United States Congress
Proposed legislationIntroduced and referred to the Senate Judiciary Committee, not law as of September 5, 2026
TAKE IT DOWN Act, Public Law 119-12
United States Congress
Enacted law or regulationEnacted as Public Law 119-12; the covered-platform notice-and-removal requirement took effect May 19, 2026
Regulation (EU) 2024/1689, Article 50
European Union
Enacted law or regulationEnacted regulation
Report on Copyright and Artificial Intelligence
UK Government
Government report or guidanceGovernment report with a separate digital-replicas consultation announced
Artificial Intelligence under the 2025 Commercials Contract
SAG-AFTRA
Collective agreementCollective agreement guidance
2026 Theatrical and Television Memorandum of Agreement
SAG-AFTRA
Collective agreementCollective agreement with some phased provisions
Interactive Digital Replicas and Consent
SAG-AFTRA
Collective agreementCollective agreement guidance
FTC Endorsement Guides: What People Are Asking
Federal Trade Commission
Government report or guidanceFederal advertising guidance
NIST AI 100-4: Reducing Risks Posed by Synthetic Content
National Institute of Standards and Technology
Government report or guidanceTechnical report
C2PA Content Credentials specification 2.4
Coalition for Content Provenance and Authenticity
Technical standardTechnical specification version 2.4

Source notes

1. Copyright and Artificial Intelligence, Part 1: Digital Replicas

U.S. Copyright Office · United States, federal · Policy report and recommendations, not enacted law

Recommends a federal digital-replica right and says licences should use informed consent, full disclosure of intended uses and limited terms.

Open the primary source

2. AB 2602: Contracts against public policy, digital replicas

California Legislature · California, United States · Enacted, applicable to specified new performances fixed on or after January 1, 2025

Makes a covered contract provision unenforceable for specified new digital-replica performances when intended use is not reasonably specific and the individual lacks the representation described by the statute.

Open the primary source

3. AB 1836: Use of likeness, digital replica

California Legislature · California, United States · Enacted

Adds liability for producing, distributing or making available certain unauthorized digital replicas of deceased personalities in expressive works, subject to listed exceptions.

Open the primary source

4. ELVIS Act, Public Chapter 588

Tennessee General Assembly · Tennessee, United States · Enacted, effective July 1, 2024

Adds voice, including a readily identifiable simulation, to Tennessee’s protected personal rights and extends liability to specified unauthorized uses and distribution.

Open the primary source

5. S.1367: NO FAKES Act of 2025

United States Congress · United States, federal · Introduced and referred to the Senate Judiciary Committee, not law as of September 5, 2026

Would create a licensable digital-replication right, require a signed writing and reasonably specific intended uses, limit duration and create civil liability for specified unauthorized activity.

Open the primary source

6. TAKE IT DOWN Act, Public Law 119-12

United States Congress · United States, federal · Enacted as Public Law 119-12; the covered-platform notice-and-removal requirement took effect May 19, 2026

Creates a narrow federal rule for nonconsensual intimate depictions, including defined digital forgeries, and requires covered platforms to operate a notice-and-removal process. It is not a general commercial likeness-licensing statute.

Open the primary source

7. Regulation (EU) 2024/1689, Article 50

European Union · European Union · Enacted regulation

Requires certain providers to mark synthetic outputs in a machine-readable and detectable format and certain deployers to disclose deepfake image, audio or video, subject to the regulation’s exceptions.

Open the primary source

8. Report on Copyright and Artificial Intelligence

UK Government · United Kingdom · Government report with a separate digital-replicas consultation announced

Says replicas can imitate a person’s voice or likeness without permission, identifies gaps in existing protection and announces further consultation rather than a completed licensing rule.

Open the primary source

9. Artificial Intelligence under the 2025 Commercials Contract

SAG-AFTRA · Covered United States commercials work · Collective agreement guidance

Describes consent for creation and use, a separate clear writing, reasonably specific intended use, added consent beyond scope, compensation, access security and storage limits.

Open the primary source

10. 2026 Theatrical and Television Memorandum of Agreement

SAG-AFTRA · Covered United States television and theatrical work · Collective agreement with some phased provisions

Addresses consent, payment, legitimate-business-need access, security, transfer obligations and use-specific treatment for digital replicas, with some terms phased to later dates.

Open the primary source

11. Interactive Digital Replicas and Consent

SAG-AFTRA · Covered United States interactive-media work · Collective agreement guidance

Describes separate written consent, a reasonably specific intended use, per-game compensation and rules allowing performers to suspend consent for generating new material during a strike.

Open the primary source

12. FTC Endorsement Guides: What People Are Asking

Federal Trade Commission · United States, federal · Federal advertising guidance

Says endorsements must be honest and not misleading and that unexpected material connections affecting evaluation should be disclosed clearly and conspicuously. It does not grant likeness permission.

Open the primary source

13. NIST AI 100-4: Reducing Risks Posed by Synthetic Content

National Institute of Standards and Technology · United States, technical guidance · Technical report

Reviews provenance, watermarking, labeling, detection, testing and auditing approaches for synthetic content. Those techniques do not establish a person’s permission by themselves.

Open the primary source

14. C2PA Content Credentials specification 2.4

Coalition for Content Provenance and Authenticity · International technical standard · Technical specification version 2.4

Defines signed, cryptographically bound assertions about an asset’s provenance and adds repository receipts plus current soft-binding and live-video mechanisms. Validation does not decide whether a depicted person consented.

Open the primary source

Reproduce the counts

  1. Download the JSON or CSV dataset.
  2. Filter each control column to true.
  3. Count the matching records. The total should agree with the chart and table above.
  4. Open each primary-source URL before relying on its status or applying it to a transaction.

Download JSON · Download CSV · Download SVG chart

Limitations and update rule

This review is intentionally selective, English-language and operational. It does not measure prevalence across all law. Source status can change after the cutoff. A later edition must retain this version, publish a new reviewed-through date, identify added or changed sources and recompute every count from the released data.

Read the reportInspect the evidence chain

Method transparency makes the counts reproducible. It does not turn this research into legal advice.

Common questions

How were sources selected?

Only primary sources from legislatures, government bodies, unions and technical standards organizations were included, and each had to directly address a control relevant to AI digital replicas.

What does addressed mean in the dataset?

It means the source expressly discusses the coded control. It does not mean the source creates a binding universal requirement.

Why include proposals and technical standards with enacted law?

They reveal different parts of the operating environment. The authority_type and status fields keep their legal weight separate.

How can the findings be reproduced?

Download the JSON or CSV, count true values for each control and verify each row against the linked primary source.