H.R. 9112 — The CREATOR Act

The CREATOR Act (Creative Rights Ensuring Artists' Technique and Originality are Reserved Act) is a proposed federal statute that would create an entirely new IP right — distinct from copyright and trademark — giving visual artists control over AI-generated outputs that deliberately impersonate their personal style for commercial purposes.

The Core Right

The bill creates a federal exclusive right for visual artists to authorize or prohibit the commercial exploitation or public distribution of a "stylistic impersonation" of their work. This is not a copyright — it doesn't protect the copying of any specific work — and it's not a trademark — it doesn't target brand confusion in the conventional sense. It sits in a newly carved lane: deliberate, deception-oriented AI style mimicry. The right lasts for the artist's lifetime, and post-mortem can be extended in five-year increments up to 50 years after death, if a notice is filed with the Copyright Office. Critically, it is assignable and licensable in writing, making it commercially transferable much like copyright.

What “Stylistic Impersonation” Actually Means

“Stylistic impersonation” requires three things to be present simultaneously: (1) the output must be generated in whole or material part by an AI system; (2) it must result from intentional configuration, prompting, design, or marketing aimed at emulating a specific artist’s distinctive visual characteristics; and (3) the resulting combination of characteristics must either mislead a reasonable viewer as to source, sponsorship, or approval, or materially affect the commercial market for the artist’s work. Evidence of intent can include explicit naming of an artist in a prompt, system configuration, or marketing materials. Critically, the definition expressly excludes general artistic influence, genre conventions, historical movements, and outputs from general-purpose AI, absent targeted intent. Style alone — in the abstract — is not protectable.

Who Is Liable and for What

The bill has two primary liability hooks. First, anyone who commercially exploits or distributes a qualifying stylistic impersonation without authorization. Second, any AI developer who both intentionally configures their system to generate stylistic impersonations of a specifically identified artist and expressly markets it as capable of doing so. The general-purpose AI carve-out is explicit and broad: mere capability to approximate a style, combined with user behavior, is not enough to trigger liability. This is an "intent plus marketing" filter designed to protect foundational model providers from strict liability based solely on what their models can technically produce.

Remedies

Plaintiffs can elect either actual damages plus the violator’s attributable profits or statutory damages. The statutory tiers range from $10,000–$100,000 per work for intentional commercial impersonation to $50,000–$150,000 per work for willful violations involving intentional targeting. Courts weigh willfulness, scale of dissemination, commercial impact, and the defendant’s compliance efforts. Defendants who establish that they fall within a safe harbor or exclusion are shielded from statutory damages entirely.

Bottom Line

The CREATOR Act is narrowly targeted and structurally conservative: it requires deliberate intent, commercial exploitation, and viewer-deception or market harm, while expressly protecting general-purpose AI, artistic influence, expressive uses, and training-data-related litigation paths. The practical exposure zone comprises AI tools and services that are explicitly marketed and configured to replicate the styles of specific living or recently deceased artists for commercial output, as well as the commercial users of such tools. General-purpose model providers operating in good faith with no artist-targeting marketing face minimal risk as drafted.

Vedros v. Sterling Group and the AI-Substitute Defense

A Pennsylvania federal court delivers a four-factor sweep against commercial-website republication — and rejects, in striking language, an AI-substitute argument that is likely to recur. The U.S. District Court for the Middle District of Pennsylvania granted summary judgment to commercial photographer Nick Vedros in his copyright infringement action against a Pennsylvania dog breeder, rejecting the defendant’s fair use defense on all four statutory factors.

The Dispute

Vedros is a commercial photographer who specializes in staged animal imagery for advertising campaigns. In 2014, he registered a copyright in a photograph, depicting a dog standing on a scale with a cat nearby. The image was created for a commercial dog-food campaign and, according to the record, involved substantial production effort. The Sterling Group of the Twin Tiers, Inc., a Labrador breeder operating a commercial website, published a blog post titled “A Breeder's Note on Canine Obesity.” Sterling placed Vedros’s photograph at the top of the article in full, without authorization, license, or attribution. The article promoted dietary recommendations and linked to products sold elsewhere on the same website. Sterling did not dispute that it had copied the photograph, and rested its case entirely on fair use under 17 U.S.C. § 107.

The Court’s Fair Use Analysis

The court found that all four statutory factors weighed against Sterling and granted Vedros’s motion in full.  The court rejected Sterling’s framing of the blog as educational, observing that the article was integrated into a for-profit business and provided at least indirect commercial benefit through links to product pages on the same website. Drawing on familiar Supreme Court and Third Circuit authority, the court emphasized that the inquiry is objective and does not turn on the defendant’s stated intent. The transformativeness analysis tracked Warhol: republishing a work without adding new meaning or purpose is not transformative, and Sterling had used the photograph for the same essential purpose that animated the original. The entire image was copied without alteration. Sterling offered no justification for using the entire photograph, and the court found the image not integral to the article’s content but, in substance, functioning as an attention-grabbing visual. The court acknowledged that the actual harm flowing from Sterling’s particular use was minimal, given the limited website traffic. It nonetheless concluded that the factor weighed against fair use, because the relevant inquiry is not the harm caused by this defendant’s isolated use but the effect on the potential market for the work if uses of this kind were widespread. Consistent with the Third Circuit’s approach to commercial, non-transformative uses, the court treated the burden as resting on Sterling to show the absence of market harm — a burden Sterling did not carry. The court then turned to the most striking argument Sterling raised: that comparable images could have been generated using artificial intelligence, and that this possibility undermined any meaningful market-harm theory. The court declined to hold that copyright withdraws its protection from works that AI could in principle have produced, treating such a holding as fundamentally incompatible with the premises of the copyright regime.

Why it matters

Three threads in Vedros deserve attention beyond the immediate holding. First, the case is a clean operationalization of Warhol in a context where the defendant is not a sophisticated cultural producer but an ordinary commercial website operator. Second, the Factor 1 reasoning matters. Sterling characterized its blog as educational, but the court treated the objective features of the use — integration into a for-profit business and indirect commercial benefit through linked product pages — as decisive, rather than Sterling’s stated editorial purpose. That has obvious implications for any defendant whose commercial purpose can be obscured by editorial wrapping.

Third, Vedros is one of the first reported decisions to engage seriously with an AI-substitutability argument as a market-harm defense — and to reject it in foundational terms. The argument has been telegraphed in defense circles for some time: if a comparable image could be generated by a text-to-image model, the plaintiff’s licensing market would be illusory, because no rational defendant would pay for a license when free substitutes exist. The court’s response was not a narrow procedural rejection but a substantive one.

A transatlantic footnote

For readers approaching this from a European perspective, a brief reminder: there is no fair use doctrine in EU copyright law. The InfoSoc Directive’s closed list of exceptions and limitations under Article 5 does not include a general fairness analysis. A Sterling-style use in the EU would be assessed under a different legal architecture. The outcome would almost certainly be the same!

Microsoft lost its copyright battle on the Windows and Office license resale

The Competition Appeal Tribunal keeps secondary markets open

The CAT ruling has substantial implications for software licensing practices in the UK and potentially broader European markets.'

The Dispute

Value Licensing (VL) operates in the business of buying and reselling pre-owned software licenses. Microsoft's position has been that the original purchaser cannot sell them on to a third party. Microsoft relied on both contractual terms (the license agreement) and copyright law (arguing resale without authorization infringes its rights) to block this secondary market.

Value Licensing argued that once Microsoft sells a copy of Windows or Office, its distribution right in that copy is "exhausted"—meaning Microsoft cannot control what happens to it afterward, including resale. This principle derives from EU law, particularly the Court of Justice's 2012 UsedSoft v Oracle decision, which established that exhaustion applies to software downloaded online, not just physical copies.

The Secondary Market Value

Secondary market licenses are significantly cheaper than buying directly from Microsoft, which threatens Microsoft's pricing power and recurring revenue model. Microsoft has historically been aggressive in shutting down resellers and warning customers against purchasing from them.  More precisely, here, VL claimed that Microsoft stifled the secondary market by requiring enterprise customers migrating to Microsoft 365 subscriptions to either surrender or retain their perpetual licenses rather than reselling them.

The Decision

The CAT addressed two preliminary copyright issues that Microsoft raised as defenses.

First, it stated that, notwithstanding contractual terms prohibiting further transfer, Microsoft can no longer oppose the resale under copyright law.

Second, Microsoft argued that, even if software rights are exhausted, its copyright in artistic works embedded in Windows and Office (user interfaces, fonts, icons, help files) cannot be exhausted. In other words, a third party cannot resell Windows and Office without Microsoft's consent. The CAT rejected this argument: where works are "ancillary" or "incidental" to computer programs sold as a unified product, exhaustion applies to the whole.

Practical Implications

This is a significant victory for the secondary software market. The ruling confirms that enterprise customers can resell portions of their perpetual licenses, and that software vendors cannot use embedded artistic works to circumvent exhaustion.

Microsoft has already obtained permission to appeal the jurisdictional ruling. This ruling represents a meaningful constraint on software vendors’ ability to use copyright to restrict secondary markets. If upheld on appeal, it may have broader implications for licensing strategies, especially where such practices intersect with competition concerns.

Different models to monetize innovations in a globalized market

The case of Japan's Shine Muscat grapes 

Japan missed the international filing window for plant-variety rights on Shine Muscat. Under the UPOV system, breeders typically have four to six years from first sale to file for protection abroad. Because Japan failed to do so, the variety can be legally grown overseas, costing an estimated ¥10 billion ($65 million) annually in lost revenue.

Limited IP Solutions

Geographical Indication protections can police misuse of place-based names, but cannot prevent cultivation of the same variety under different branding elsewhere. In other words, once plant material escapes unlicensed cultivation, it proliferates regardless of Japan's preferences. Trademarks cannot help either: Japanese law explicitly bars registering plant variety names to prevent monopolization of varietal terms. With the foreign filing window closed, plant variety rights cannot be reasserted overseas, leaving Japan with limited recourse against existing unauthorized cultivation.

The Other options: Export Model and  Licensing Model

The traditional export model approach is straightforward: grow premium products domestically and ship them abroad. Japan's $30 Shine Muscat grapes exemplify this—the high price reflects Japanese terroir, quality control, and brand cachet. Domestic producers understandably prefer this model because it preserves their competitive position and captures the whole value chain domestically. The goal is to capture value through scarcity and quality differentiation rather than through IP fees. The problem: perishability limits reach.

The alternative is contractual protection. Under the licensing model, Japan would authorize vetted foreign producers to cultivate its proprietary varieties under contractual terms, collect royalties, maintain quality standards, and create a year-round supply that Japanese seasons can't provide. These agreements specify quality standards the licensee must maintain, geographic or market restrictions on where the product can be sold, branding requirements, and audit rights to ensure compliance. This approach also addresses the enforcement reality: managed licensing recaptures some value from inevitable diffusion (Shine Muscat leaked abroad to China and Korea).

The Tension

Licensing may generate sustainable long-term revenue streams and brand control, but it also cultivates competitors who may eventually outcompete producers on cost.  The knowledge doesn't stay contained, and domestic growers' premium positioning erodes.

Under a pure export model, any knowledge that escapes is incidental leakage rather than a deliberate revenue strategy. Japan didn't intend for Shine Muscat vines to reach Chinese and Korean growers—that was unauthorized propagation that exploited the missed UPOV filing window. The knowledge sharing happened, but Japan captured nothing from it. This is happening while food has become central to Japan's "Cool Japan" soft-power strategy, with agricultural exports explicitly listed as a pillar with 2033 expansion goals tied directly to IP policy, and Japan continues expanding mutual-recognition agreements with partners like the European Union.

Interestingly, this case mirrors debates in other IP-intensive industries. Pharmaceutical companies face similar choices about licensing generic production versus defending exclusivity. Luxury brands wrestle with whether controlled expansion causes dilution.