As of Q3 2026, the legal implications of AI resurrected actors are no longer confined to state publicity claims, guild bargaining, or platform moderation judgment calls. The NO FAKES Act, reintroduced on May 20, 2026 as S.4591/H.R.8915, cleared the Senate Judiciary Committee unanimously on June 18, 2026, moving it from another federal AI-rights proposal into the narrower category of legislation that counsel may need to operationalize on short notice.[1]
That status matters, but so does the discipline around it. The bill is still proposed legislation. It may be amended before any floor vote, and passage is not guaranteed. California, Tennessee, and other state frameworks remain the enacted law practitioners must deal with today. The constitutional critique of the bill, including the Reno v. ACLU and Ashcroft v. ACLU overbreadth framing discussed below, is academic analysis rather than a judicial holding on this bill. The useful question is therefore not whether federal law has already solved the digital-replica problem. It has not. The question is what this bill would require if enacted in substantially similar form.
The Bill Is About More Than Celebrity Publicity
The most consequential move in the 2026 version is its scope. The bill would create a new federal intellectual property right in an individual's voice and visual likeness, and that right would belong to every individual, not only to famous performers or estates with recognizable commercial value.[2]
That choice changes the compliance problem. A resurrected movie star in a studio trailer is the clean headline, but the bill is not limited to dead celebrities, theatrical releases, or commercial endorsements. It reaches the creation and distribution of unauthorized AI-generated digital replicas of voice or likeness under a federal structure. For platforms and entertainment companies, the right would have to be administered at scale, across disputed identities, disputed authorization, and disputed expressive purpose.
The state-law background explains why Congress is here, but it should not be overstated. State publicity and digital replica laws have produced a patchwork that already matters to performers, studios, heirs, and online services. That patchwork problem is visible in performer-rights disputes such as the issues discussed in Sandra Bullock AI comments, but the federal bill would not simply erase that environment. It would sit on top of some existing state protections while cutting off others.
The Operating Mechanism: Notice, Takedown, Counter-Notice
The bill's practical center is a DMCA-style notice-and-takedown regime. A rights holder or authorized representative would submit a notice alleging an unauthorized digital replica. The online service would then face a procedural choice: remove or disable access, evaluate whether the notice is facially sufficient, and account for the possibility that the uploader or speaker will challenge the claim through a counter-notification process.[2]
The counter-notification feature is not decoration. It is the procedural hinge that keeps the system from becoming a one-way deletion channel, but it also creates new legal exposure. Under the 2026 version described by bill analysis, a person who knowingly files a false counter-notification may face $25,000 per instance or actual damages plus fees.[2]
That number will change behavior. A documentary producer, parody account, journalist, estate representative, or fan creator might all believe they are on legally defensible ground. The penalty means the counter-notice is not merely a moderation appeal. It is a representation made under threat of statutory consequences. Counsel reviewing one will need to ask not only whether an exception might apply, but whether the record is strong enough to let a client make that assertion without turning a speech dispute into a damages dispute.
| Actor | Likely decision point if the bill passes |
|---|---|
| Performer, estate, or authorized representative | Whether the replica sufficiently uses protected voice or likeness and whether the notice can be made accurately |
| Platform or hosting service | Whether the notice is facially valid, whether to remove or disable access, and how to handle a counter-notification |
| Uploader, producer, or speaker | Whether to accept removal, revise the work, seek authorization, or file a counter-notification despite penalty exposure |
| Studio or distributor | Whether existing releases, promotional assets, and vendor-generated materials have the consents needed under a federal digital replica right |
| Counsel | Whether the work falls within an exception, whether state law still applies, and whether litigation risk justifies keeping the material online |
For in-house teams, the immediate burden would be less glamorous than the subject matter. Intake forms must capture identity, authorization, chain of rights, the allegedly replicated voice or likeness, and the location of the challenged material. Platform policy teams would need escalation routes for disputes involving news, parody, documentary, scholarship, and education. Entertainment counsel would need to align clearance files with public-facing takedown positions; an authorization clause buried in a production agreement will not help if the platform reviewer cannot verify it quickly.
Contractual AI controls already appearing in studio and streamer practices would not become irrelevant. They would become part of the evidentiary record. The production-side controls discussed in Netflix AI production rules illustrate why internal approval logs, performer consents, vendor obligations, and edit histories may matter when a federal notice arrives.
Where the Workflow Runs Into Speech
The bill attempts to avoid obvious First Amendment collisions through carve-outs. Reported exclusions include news, parody, criticism, documentary, scholarship, and, in the 2026 iteration, nonprofit libraries, archives, and educational institutions.[3]
Those carve-outs do real work. A law without them would be far easier to administer and far harder to defend. But exceptions are also where lawyers will spend their time. A political satire, a biographical documentary, a classroom archive, a news segment using a synthetic reenactment, and a commercial trailer can all involve a recognizable voice or likeness. The legal question is not whether the technology is AI. It is whether the use falls inside the federal right, outside it, or in the unstable middle where a notice is plausible and a counter-notice is dangerous.
The parody problem is especially likely to resist clean processing. The issues raised in Al Pacino AI parody analysis show why recognizable imitation can be both the substance of the speech and the basis of the rights claim. A takedown workflow will push that ambiguity into deadlines, templates, and risk thresholds.
The constitutional concern follows from the machinery, not from a generalized discomfort with regulating deepfakes. Academic analysis has argued that the bill's broad restrictions on online speech may face overbreadth challenges under Reno v. ACLU and Ashcroft v. ACLU, and has noted that the current administration has signaled skepticism toward restrictions that could chill protected speech.[3]
That does not mean the bill is constitutionally doomed. Reno and Ashcroft are being used here as an analytical frame for online-speech breadth and chilling effects, not as a direct prediction of judicial invalidation. The harder point is operational: a platform or studio cannot wait for a court to settle the outer boundary before it responds to notices. It must decide whether material stays up, comes down, or moves into a counter-notice path while the constitutional status of the overall scheme remains contestable.
Preemption Narrows the Patchwork, but Does Not Erase It
The 2026 bill's preemption compromise is one of its most lawyerly features. It would preserve state laws enacted before January 2, 2025, including frameworks such as California AB 1836 and Tennessee's ELVIS Act, while preempting future state digital replica legislation.[1]
That approach answers one concern while creating another. It prevents a rolling fifty-state race to enact new digital-replica statutes after federal passage. At the same time, it preserves older state-law regimes that may differ in scope, remedies, descendibility, consent rules, or procedural posture. A federal right would not necessarily give counsel a single governing answer.
Practitioners should be careful with dates here. Reported effective dates for California AB 1836 and AB 2602 differ across sources, with some citing January 1, 2025 and others January 1, 2026. Official California legislative text should be checked before making any preemption conclusion turn on those dates. New York also requires precision: the newer 2025 digital replica framework associated with SAG-AFTRA-backed bills is distinct from New York's earlier 2021 right of publicity statute.
The federal-state line is therefore not an academic footnote. A claimant may have a federal notice theory, a preserved state-law theory, or both. A platform may be able to process the federal notice while still facing state-law exposure outside the takedown channel. A studio may have cleared a performer under contract and still need to assess whether postmortem or replica-specific state rights create a separate problem.
What Counsel Should Prepare to Parse
If the NO FAKES Act moves toward a floor vote before the end of Q3 2026, the compressed political calendar will matter. Senator Thom Tillis remains one of the Senate's key intellectual-property voices, and his retirement at the end of 2026 gives the legislation a narrower window than an ordinary open-ended reform project. That timing issue sits alongside the broader political analysis of Tillis's IP role in Thom Tillis political analysis.
The first preparation task is notice quality. Rights holders should not assume that an emotionally obvious misuse will translate into a legally sufficient federal notice. The notice must identify the right, the allegedly unauthorized replica, the claimant's authority, and the location of the material in a way a service can process. Estates and representatives will need documentation ready before the disputed clip goes viral.
The second task is exception review. Platforms should avoid treating categories such as parody, criticism, documentary, and scholarship as magic words that automatically defeat a notice. Claimants should likewise avoid assuming that a monetized or widely distributed work falls outside expressive protection. The carve-out analysis will often depend on the work's function, context, and relationship to the person being replicated.
The third task is counter-notification discipline. A counter-notice may be the right answer when a takedown is overreaching, especially in news, criticism, or parody contexts. But the $25,000 false-counter-notification exposure means counsel should treat it as a legal filing, not as a customer-service response.[2]
The fourth task is state-law mapping. Before assuming the federal bill simplifies everything, counsel should identify which preserved state regimes may still apply, which future state digital replica claims would be preempted, and whether older publicity laws reach the same conduct through a different theory.
The fifth task is constitutional risk allocation. Companies will still need policy rules for the period between enactment and judicial interpretation. Some will remove aggressively to avoid liability. Others, particularly media and commentary platforms, may preserve more disputed material where the expressive claim is strong. Neither choice is free. Over-removal chills speech; under-removal invites statutory claims.
If enacted in substantially similar form, the NO FAKES Act would be the most credible federal digital replica right yet. Its practical value will depend less on whether lawyers can describe the bill in broad performer-rights terms and more on whether they can parse notices, exceptions, counter-notices, surviving state laws, and constitutional uncertainty quickly enough for the platform and production timelines in which these disputes actually arise.
References
- Senate Judiciary Committee Advances Legislation to Protect Name, Image and Likeness Rights from Unauthorized AI Use, HK Law, June 2026.
- Congress Reintroduces the NO FAKES Act: What's New in the 2026 Bill, Manatt.
- A New Age of Publicity: The NO FAKES Act and Federal Regulation on AI Replicas, CU Law Review.