NO FAKES Act: Four Stakeholders, One Bill, and What the Senate Path Actually Looks Like

A unanimous committee vote is a signal, not a guarantee. The NO FAKES Act has cleared its first major legislative hurdle. What it faces next is more complicated than the vote count suggests.

What the Bill Actually Does

S. 4591, formally the Nurture Originals, Foster Art, and Keep Entertainment Safe Act of 2026, would create a new federal intellectual property right giving every individual the right to control how their voice and visual likeness appear in AI-generated digital replicas. Every individual. Not just celebrities. Not just public figures. The person on a corporate training video, the call center operator whose voice was recorded, the background actor whose image appeared in a dataset, each of them would hold a federal IP right that didn’t exist before.

The bill holds anyone who produces or distributes an unauthorized digital replica liable under that right. Companies, individuals, and platforms sit inside that liability perimeter. Reported platform penalties reach up to $750,000 per unauthorized work, according to coverage of the bill text. That specific figure wasn’t confirmed in the accessible legal analysis reviewed for this brief, treat it as reported, not established, until the bill text is directly accessible. What the Holland & Knight alert does confirm is that civil liability exists for creators and distributors, and the framework covers both visual and audio likeness.

The bill includes First Amendment exemptions, reportedly covering news reporting, parody, satire, criticism, and documentary use. Those exemptions are where the contested ground begins.

The Stakeholder Map

Four positions are on the record. Understanding them is the compliance team’s job before the floor vote.

*Bill sponsors.* The bipartisan sponsorship, which enabled the unanimous committee vote, reflects a coalition built around talent protections and creative industry rights. The entertainment industry’s concern about AI-generated synthetic performances driving down the market for licensed human talent is the political engine behind this bill. The sponsors’ position is that the federal IP right is a floor, not a ceiling: it establishes a baseline protection that state right-of-publicity laws have provided inconsistently.

*Platform industry.* No platform trade group has been quoted directly in the accessible sources for as of publication. But the platform compliance question is straightforward: any operator hosting or distributing AI-generated content that could constitute a digital replica of a real person’s voice or likeness faces exposure under the liability framework. The bill’s definition of “digital replica” and the scope of “authorization” are the two variables the platform industry will push hardest on during floor amendment negotiations.

*Public Knowledge.* The organization has specifically warned that the bill risks enabling misuse of individuals’ likenesses through the very consent mechanism meant to protect them. The specific concern: click-wrap contracts could allow individuals to be locked into licensing their likeness rights for extended periods, reportedly as long as ten years in some readings of the draft, without meaningful informed consent. That’s a warning about the bill’s consent architecture, not its intent. The distinction matters for floor amendment strategy.

*Electronic Frontier Foundation.* The EFF has raised concerns that the bill’s First Amendment exemptions may be insufficient to protect satire, commentary, and news reporting in practice. The EFF source for this brief was inaccessible, so this position is attributed rather than confirmed from the specific June 2026 analysis. The concern is structurally consistent with EFF’s prior positions on digital IP legislation and is worth tracking as the floor debate develops.

These four positions aren’t symmetrical. Bill sponsors and the talent industry have the political momentum, a unanimous committee vote proves it. The civil liberties concerns are substantively serious but haven’t generated enough political friction to slow the bill at committee stage. The platform industry’s position will be expressed through amendment pressure, not opposition votes.

The Legislative Path

The Senate Judiciary Committee vote on June 18 was one procedural step. Here’s the remaining sequence.

Full Senate floor vote is next. The Senate calendar is contested. The bill could move relatively quickly if leadership prioritizes it, or it could wait behind budget and appropriations business. No floor vote date has been announced. The unanimous committee vote removes the risk of a committee hold, but a floor hold from any individual senator remains possible.

The House companion bill hasn’t been marked up by the House Judiciary Committee. That gap is meaningful. Even if the full Senate passes S. 4591, the House needs to pass a companion version, and if the two versions differ, a conference reconciliation process follows. TJS covered the NO FAKES Act safe harbor dynamics earlier this year; the federal-versus-state preemption question flagged in that coverage becomes acute at the conference stage, where House members from states with strong right-of-publicity laws will have opinions about how the federal framework interacts with state statutes.

Realistically: if everything moves efficiently, enactment is a late 2026 scenario. A more measured estimate puts the effective date in 2027, with implementation guidance following six to twelve months after enactment. Compliance teams that start impact assessment now have roughly 12 to 24 months to build consent infrastructure, audit training datasets for likeness data, and revise terms of service, assuming the bill passes in something close to its current form.

What the Bill Would Require of AI Content Operators

The compliance question is operational, not theoretical. Under the bill’s framework as currently described, an AI content platform would need to:

Establish authorization records for any digital replica of a real person’s voice or visual likeness it produces or distributes. That means consent documentation that can demonstrate the individual knew what they were authorizing, for what purpose, and for how long. If the click-wrap concern raised by Public Knowledge is addressed in a floor amendment, by requiring clearer disclosure or limiting contract duration, the consent infrastructure requirements become more demanding, not less.

Audit existing training data and generated content for coverage. The bill’s liability framework applies to unauthorized replicas. A platform that trained on datasets containing voice or likeness data without consent documentation faces retroactive exposure if the dataset included individuals who didn’t meaningfully authorize their inclusion. The definition of “authorization” in the final text will determine how extensive this audit obligation is.

Review First Amendment reliance. Satire and parody protections are in the bill, reportedly. But the adequacy of those protections, the question the EFF raised, will be litigated. A platform that relies on a fair use or satire defense without confirmed statutory clarity is taking a risk the bill’s current draft hasn’t resolved.

What Remains Unresolved

Three questions will define the compliance picture and aren’t answered by the committee vote.

The $750,000 per-work penalty figure: consistent with prior reporting on the bill’s structure, but not confirmed in the accessible legal analysis reviewed here. If accurate, this number changes the risk calculus for platforms with large libraries of AI-generated content substantially. A platform hosting thousands of items would face aggregate exposure that makes individual claim settlement economics look very different from what state right-of-publicity litigation has historically produced.

The First Amendment exemption scope: the specific categories reportedly covered, news reporting, parody, satire, criticism, documentary, need confirmation against the bill text. The EFF’s concern suggests that as drafted, the exemptions may be narrower in practice than the statutory language implies. This is the most likely target for civil liberties-focused floor amendments.

The click-wrap problem: Public Knowledge’s concern about extended licensing lock-in through click-wrap contracts is a consent architecture problem, not a drafting error. Fixing it requires either a durational limit on likeness licensing through click-wrap, an affirmative disclosure requirement, or both. Whether that fix appears as a floor amendment or in conference is unknown. The federal-versus-state compliance map for digital replica law becomes more complex, not simpler, if the federal framework passes with weak consent protections that states then attempt to supplement.

TJS Synthesis

The unanimous committee vote is the most important thing that happened to AI content regulation in the United States in June 2026. Not because the bill is law, it isn’t, but because it removed the question of whether a federal digital replica framework could attract bipartisan support. It can, and it did.

The compliance posture should shift accordingly. Building an AI compliance program for a patchwork landscape means accounting for the NO FAKES Act as a near-term probability rather than a distant possibility. The specific penalty figures, exemption scope, and consent architecture will be shaped by floor amendments and conference reconciliation. But the core architecture, a federal IP right over digital likeness, with civil liability for unauthorized replicas, is what the committee unanimously approved.

What comes next: floor vote timing and the first House markup hearing. Those two events will determine whether the NO FAKES Act becomes law in 2026 or early 2027. The compliance window is open. It won’t stay open indefinitely.