NO-FAKES Act 48-Hour Notice Compliance: What Must Platforms and Trademark Holders Do?

What the NO-FAKES Act Actually Requires Within 48 Hours

The Nurture Originals, Foster Art, and Keep Entertainment Safe Act (NO-FAKES Act) imposes a statutory duty on covered online platforms to act on a properly submitted notice of an unauthorized digital replica of an individual's voice or visual likeness within 48 hours. The bill, which was reported to be headed for a Senate floor vote in June 2025, frames that 48-hour window as the maximum period between receipt of a valid notice and either the removal of, or the disabling of access to, the implicated content. Failure to comply exposes a platform to civil liability of up to $750,000 per violation, layered on top of any existing right of publicity, copyright, or trademark claim that the notice-giver might pursue separately. For trademark practitioners, the 48-hour clock is the most operationally concrete number in the entire statute, because every other provision is contingent on what counts as a "valid notice" and what counts as "expeditious" action after one is received. As of the September 2026 date context, the Senate has not yet passed a final bill, but the language circulating in committee prints and the parallel implementation timeline of the TAKE IT DOWN Act (which went live in 2025 with its own notice-and-takedown regime) make the 48-hour standard the de facto planning figure for counsel advising brand owners.

Also worth reading: What is the AI trademark compliance checklist for verifying generated brand assets? · What is the current USPTO AI filing guidance compliance standard for trademark and patent filings in 2026? · What are the essential AI trademark compliance strategies for 2027?

The 48-hour obligation is not a generic "best efforts" duty. It is paired with a safe harbor: a platform that acts within 48 hours, retains the removed content for at least 90 days for evidentiary purposes, and notifies the uploader of the action taken cannot be held liable for the removal itself. That safe-harbar architecture is borrowed almost verbatim from section 230 of the Communications Decency Act and from the takedown provisions of the Digital Millennium Copyright Act, but it is sharper-edged than either because of the explicit numeric deadline. Critics have noted that 48 hours is shorter than the typical 5-to-10 business day cycle of a brand-protection vendor's first manual review, and shorter than the 24-hour clock imposed under the European Union's Digital Services Act for certain categories of "manifestly illegal" content. The result is a regime in which a trademark or right-of-publicity holder can, in effect, obtain a removal order against a deepfake of their mark, spokesperson, or product endorsement without filing suit.

How the 48-Hour Clock Interacts with Trademark Claims

Trademark owners often assume that the NO-FAKES Act is a right-of-publicity statute and therefore irrelevant to brands. That is wrong in two specific ways. First, the bill's definition of "digital replica" includes not only the voice and likeness of a natural person but also "a performer's distinctive voice, performance, or likeness" where the performer is identifiable to a reasonable consumer. Many brand mascots, AI-generated brand ambassadors, and synthetic spokesperson voices that have acquired secondary meaning in commerce can fall within that definition if they are tied to a living or recently deceased identifiable individual. Second, the Act expressly preserves all other rights and remedies, including those under the Lanham Act. A trademark holder whose mark appears in a deepfake advertisement for counterfeit goods can therefore use the NO-FAKES 48-hour channel to obtain the takedown and then layer a traditional Lanham Act false-advertising, dilution-by-blurring, or counterfeiting suit on top of it. The 48-hour removal does not extinguish the Lanham claim; it just removes the offending asset before consumer confusion compounds.

The interplay matters for proof. Because the platform must retain the removed content for at least 90 days, the trademark owner should send a litigation hold request to the platform within that window. If the platform fails to preserve the content, the safe harbor may not protect it, and the loss of evidence can be used as an adverse inference instruction in subsequent trademark litigation. Practitioners advising brands that rely on synthetic spokespersons or AI voice clones should treat every 48-hour removal as the opening move of a multi-stage evidentiary process, not as a one-shot fix.

A Practical 48-Hour Compliance Workflow for Brand Owners

The most defensible workflow combines automated watch services with a short, content-rich notice template. First, a brand-protection platform (such as those operated by the major corporate trademark service providers, plus boutique AI-scrape vendors) flags a suspected unauthorized digital replica within hours of publication. Second, counsel prepares a NO-FAKES-compliant notice that includes: (1) the trademark registration number or pending application serial number for any asserted mark; (2) the URL or persistent identifier of the infringing content; (3) a sworn or signed statement that the content is an unauthorized digital replica; (4) contact information for the rights holder; and (6) a good-faith statement under penalty of perjury. Third, the notice is submitted through the platform's designated agent, which the statute requires platforms to designate in a manner similar to the DMCA's designated agent registry under 17 U.S.C. § 512(c). Fourth, counsel logs the timestamp of receipt; if the platform has not removed or disabled access within 48 hours, counsel sends a second notice referencing the first, citing the $750,000 per-violation statutory damages figure, and reserving rights.

The workflow fails when notices are sent to general legal inboxes rather than to the designated agent, when the notice omits the trademark registration evidence, or when the brand sends a single form-letter notice for hundreds of URLs without itemizing each. Platforms have publicly resisted batched notices in their comments on parallel rulemakings, and the TAKE IT DOWN Act's enforcement record through 2025 shows that the Federal Trade Commission has brought actions against platforms whose intake systems collapsed under undifferentiated bulk submissions. Sending a clean, narrow, evidence-rich notice is therefore both faster and more durable.

Comparison: NO-FAKES vs. TAKE IT DOWN vs. DMCA Notice-and-Takedown

The three regimes are similar in shape but differ in trigger, deadline, and remedy. The table below compares the operative provisions as understood from public drafts and from the TAKE IT DOWN Act's live enforcement record as of late 2025.

The comparison shows that the NO-FAKES Act is best understood as the strictest of the three, in deadline and in penalty, while borrowing the procedural shape of the DMCA. For trademark counsel, the practical takeaway is that NO-FAKES can be used for replica-related takedowns faster than the DMCA, and that DMCA remains the correct vehicle for purely textual or pictorial copying of a mark that does not involve a person's voice or likeness.

Common Mistakes That Defeat a NO-FAKES Notice

The most common mistake is treating the notice as a cease-and-desist letter. A cease-and-desist is a private communication demanding voluntary action; a NO-FAKES notice is a statutory mechanism that starts a 48-hour clock and unlocks statutory damages. Sending a "soft" letter that asks the platform to "consider" removal resets nothing because the clock does not start until a notice meeting the statutory criteria is received. A second common mistake is failing to identify the rights holder as a natural person whose voice or likeness is at issue. The bill, as drafted, is person-centric. A corporation whose only asset at issue is a logo cannot use NO-FAKES; it must use the Lanham Act and the DMCA. A third mistake is underestimating the platform's counter-notice rights. The proposed text allows the uploader to submit a counter-notice, after which the platform may restore the content unless the rights holder files a federal action within a defined window (commonly 10 to 14 business days in parallel regimes). Brand owners who rely on a NO-FAKES removal as their only enforcement step can find the content reposted on day 15.

A fourth mistake is ignoring the 90-day retention rule. Once the platform removes the content, counsel has 90 days to obtain a forensic copy, including metadata, hash values, and account-level identifiers. Missing that window forfeits the most valuable piece of evidence a brand owner has in any subsequent trademark infringement suit, because deepfake creators typically rotate accounts and re-upload under new hashes within weeks. A final mistake is conflating state right-of-publicity statutes with the federal regime. Roughly 27 states have right-of-publicity statutes, and several (notably California, Tennessee, and Indiana) recognize post-mortem rights of 50 to 100 years. The federal Act preempts conflicting state law on the 48-hour takedown mechanism but explicitly preserves the underlying state-law rights of publicity, which means a brand owner should plead both state and federal theories in any complaint filed after a non-removal.

When to Act and What It Costs

The 48-hour clock is short enough that a brand owner cannot wait for a quarterly watch report. Counsel should treat a confirmed unauthorized replica as a same-day task. The marginal cost of a NO-FAKES notice is low: many brand-protection platforms include notice-drafting as part of their subscription, and outside counsel typically bills 1.5 to 3 hours to prepare and submit a single, well-evidenced notice, with each subsequent notice on a related platform taking less time. The cost of failing to act is asymmetric. A single 48-hour period in which a deepfake of a brand ambassador endorses a competing product can move material consumer-attribution data, and the $750,000 statutory damages figure per violation is materially higher than the typical $30,000 to $150,000 statutory award under the Lanham Act's 15 U.S.C. § 1117(d) provision for counterfeiting, although trademark remedies can be trebled for willful infringement. From a portfolio perspective, the most efficient posture is to maintain a pre-drafted notice template, a current list of designated agents for the top social and generative-AI platforms, and a documented chain of custody for any forensic image of the offending replica captured the moment it is detected.

Critical and Nuanced Takeaways

The 48-hour number is real, but the statute is not yet law in the form most commentators describe. As of September 2026, the Senate has held hearings and committee votes, but the bill's enactment, signature, and effective-date language remain uncertain. The TAKE IT DOWN Act's effective-date history, which saw the FTC publish its enforcement priorities within roughly 90 days of enactment, is a useful proxy for how quickly a NO-FAKES safe-harbar framework could come online. Until the statute is enacted, brand owners cannot rely on the 48-hour clock as a matter of law, but they can — and several major brands already do — use the NO-FAKES-style notice format as a matter of contract with platforms that have voluntarily adopted it through updated terms of service. That voluntary adoption is uneven: some platforms accept the notice format, others reject it as premature, and at least one major social network has publicly disputed the FTC's enforcement priorities under the parallel TAKE IT DOWN Act.

There is also a quieter trademark question that the bill does not resolve: whether the use of a mark in an AI-generated replica of a person constitutes "use in commerce" under 15 U.S.C. § 1127. If the replica is an endorsement and the replica's sponsor receives advertising revenue, the answer is likely yes; if the replica is purely satirical or editorial, the First Amendment and the Rogers v. Grimaldi balancing test (applied recently in the Jack Daniel's v. VIP Products Supreme Court decision) push toward non-liability for the platform. Counsel should therefore plead trademark claims in the alternative rather than as the lead theory. The NO-FAKES 48-hour channel is best used as the fastest route to evidence preservation and to consumer-confusion mitigation, while the substantive Lanham Act claim remains the path to damages.

Conclusion

The NO-FAKES Act's 48-hour notice-and-takedown requirement is the operational pivot of the statute, sitting between a rights-holder's first detection of an unauthorized digital replica and the platform's statutory exposure to $750,000 per violation. For trademark owners, the channel is most useful when the replica implicates an identifiable voice or likeness tied to a brand asset and when the brand is prepared to preserve evidence within the 90-day retention window. The biggest avoidable losses come from misformatted notices, from ignoring the counter-notice window, and from treating the 48-hour removal as a substitute for, rather than a complement to, a federal Lanham Act or state right-of-publicity claim. As the bill progresses and as platforms codify their intake procedures, the brands that have built pre-drafted notice templates, current designated-agent lists, and same-day forensic capture processes will find the 48-hour clock to be the shortest distance between detection and deterrence.