NO FAKES Act Impact Assessment: What a Federal Voice and Likeness Right Would Require of Platforms and AI Devs

BILL STATUS: Passed Senate Judiciary Committee, Not Yet Law.

The bill is real. The compliance obligations aren’t, yet.

That distinction is doing a lot of work right now, and getting it wrong in either direction creates problems. Treating the NO FAKES Act as enacted law causes organizations to over-invest in compliance workflows for requirements that may still change during Senate floor consideration. Dismissing it as “just a committee vote” causes organizations to miss the preparation window that only exists while the bill is still moving through the process.

The right posture is a third option: understand what the bill requires in its current reported form, assess your exposure against those provisions, and position now to respond quickly when the Senate floor vote comes.

The Legislative Event: Where the NO FAKES Act Actually Stands

The Senate Judiciary Committee unanimously advanced the NO FAKES Act. Introduced by Senators Chris Coons (D-Del.) and Marsha Blackburn (R-Tenn.), along with House sponsors Representatives Madeleine Dean (D-PA) and María Elvira Salazar (R-FL), the bill has navigated the committee phase, typically the most politically complex stage for legislation with competing commercial and creative-economy interests, with enough consensus to move forward.

Committee passage is one of five steps between a bill’s introduction and enactment. The sequence from here: Senate floor vote → House introduction and passage → Conference committee (if versions differ) → Presidential signature. Each step takes time. Each step is also a point where provisions can be amended. The $750,000 penalty figure that appears in current coverage is from the bill as reported, not the bill as enacted. It can change. The 70-year post-death term cited in some coverage requires verification against the current bill text before it’s used as a compliance planning input.

What’s unlikely to change is the bill’s core structure: a federal IP right over voice and visual likeness, enforced through a platform-obligation mechanism. That architecture has survived committee and reflects the bill’s bipartisan political coalition. Organizations should build compliance assessments around that architecture while treating specific figures as provisional.

The Federal IP Right: What It Creates and Why It’s Different From State Laws

Currently, voice and likeness protection in the US is a state-law patchwork. California has a right of publicity. New York has one. A growing number of states have enacted or are considering AI-specific synthetic content laws. They don’t align. They don’t preempt each other. And they create fragmented compliance obligations for any platform or AI developer operating nationally.

The NO FAKES Act would create a federal intellectual property right protecting individuals’ voice and visual likeness from unauthorized AI-generated replicas. Per the bill’s sponsors, the right would extend to families of individuals after they pass away, placing it in the same conceptual category as copyright for original creative works. That specific duration provision requires bill text confirmation before it’s used as a compliance planning input, but it signals that Congress is treating voice and likeness as long-duration intellectual property rather than a personal right that expires at death.

For AI developers, the creation of a federal IP right has a structural implication beyond the obvious: it establishes that voice and likeness are protectable property, not just personal dignity interests. That’s a meaningful distinction in litigation, in licensing negotiations, and in how training data agreements get drafted going forward.

The Notice-and-Takedown Mechanism: What Platforms Would Face

The enforcement architecture is where the compliance burden lands most concretely. The bill includes a DMCA-style notice-and-takedown process for unauthorized digital replicas, meaning platforms that host AI-generated content would operate under a structured obligation to respond to removal notices, maintain compliant notice-receipt workflows, and avoid liability through timely takedown.

The DMCA analogy is instructive and imperfect. DMCA safe harbor applies to user-uploaded content. The NO FAKES Act’s mechanism, if it mirrors that architecture, would extend the platform obligation to AI-generated outputs, content the platform or its AI tools created, not just content users uploaded. That’s a broader operational scope than most DMCA compliance programs currently cover.

Platforms with existing DMCA workflows have a head start, not a solved problem. The workflow extensions required to cover AI-generated replica claims are real. Organizations that have assumed their DMCA compliance program covers AI-generated content should verify that assumption before the Senate floor vote.

The Penalty Structure: What $750,000 Per Violation Would Mean

The bill would expose online platforms to statutory damages of up to $750,000 per unauthorized AI-generated replica for platforms that haven’t undertaken a good faith effort to comply with the notice-and-takedown provisions. That figure requires verification against the current bill text before it’s used in risk modeling. Do not put $750,000 in a risk register as a confirmed exposure figure, it may change before enactment.

What the figure illustrates, even as a reported provisional number, is the intended deterrence architecture. Statutory damages at that level mean that the liability calculation for a platform hosting AI-generated replica content isn’t about actual harm to the individual, it’s about the number of replicas. A platform hosting AI-generated content at scale faces a multiplication problem. Whether the final enacted figure is $750,000 or a different amount, the structure is punitive-per-occurrence, not compensatory-per-harm. That structural feature will survive legislative adjustments even if the specific number changes.

Compliance Positioning Now: What to Do Before the Senate Floor Vote

Four actions are available today that don’t depend on the bill’s final text.

First, inventory. Identify every AI-generated content workflow in your organization that involves voice synthesis, facial synthesis, or visual likeness replication. This inventory is necessary for DMCA compliance already in many jurisdictions; the NO FAKES Act adds federal scope to an obligation that exists in some form at the state level today.

Second, gap-assess your notice-and-takedown workflows. If your current DMCA workflow doesn’t cover AI-generated outputs, content your tools generated rather than content users uploaded, document that gap now. Closing it after enactment under time pressure is more expensive than addressing it in the preparation window.

Third, verify the bill text. The $750,000 figure and the post-death duration term need to be confirmed against the actual legislative record before they go into compliance risk models. Congress.gov and the Senate Judiciary Committee’s official release are the appropriate sources.

Fourth, track the state law map in parallel. The NO FAKES Act’s preemption scope, what it does and doesn’t displace in state right-of-publicity regimes, will determine whether federal compliance replaces or layers on top of state obligations. That preemption question is unresolved in the current reported bill provisions. California’s AB 1836 and New York’s likeness laws don’t automatically go away on federal enactment.

The real question isn’t whether the NO FAKES Act will create new compliance obligations for platforms and AI developers. It will. The question is how much preparation time remains between now and enactment, and whether your organization uses that window or doesn’t. Committee advancement means the preparation window is measurably shorter than it was before the vote.