Neuropublic S.A., a Greek technology company, has filed a federal lawsuit against the law firm Ladas & Parry LLP, with several claims stemming from the firm’s alleged mishandling of Neuropublic’s confidential invention disclosure — sending it out to a third party (“PatentManiac”) for a preliminary novelty search which then (again allegedly) further leaked the disclosure. Although the case does not involve submission to AI algorithms, some of the questions here are similar to those many IP attorneys are considering when onboarding new AI tools. (more…)
The pending Hearst v. Martinelli case may be the “dead letter” offered by Justice Gorsuch. This time, the Supreme Court might actually decide whether the “discovery rule” applies to the Copyright Act’s statute of limitations.
On May 9, 2024, the Supreme Court issued its opinion in Warner Chappell Music v. Nealy, No. 22-1078, 601 U.S. ___ (2024), resolving a circuit split over the availability of back-damages in copyright infringement cases. In a 6-3 decision authored by Justice Kagan, the Court affirmed the Eleventh Circuit’s ruling, permitting recovery of damages for acts that occurred more than three years before the filing of the lawsuit under the “discovery accrual rule.”
For those of you who have not been following the case, the Plaintiff Sherman Nealy, a music producer, helped create musical works in the 1980s with his collaborator, Tony Butler. Nealy was incarcerated from 1989 to 2008 and again from 2012 to 2015 (more…)
This may be a useful case for patent prosecutors to cite to the USPTO because it creates a strong dividing line for the printed matter doctrine — applying the doctrine only to cases where the claims recite the communicative content of information.
In this decision, the Federal Circuit partially reversed a PTAB invalidity finding against several IOEngine patent claims. The most interesting portion of the opinion focuses on the printed matter doctrine. Under the doctrine, certain “printed matter” is given no patentable weight because it is deemed to fall outside the scope of patentable subject matter. C R Bard Inc. v. AngioDynamics, Inc., 979 F.3d 1372 (Fed. Cir. 2020). In this case though the Federal Circuit concluded that the Board erred in giving no weight to IOEngine’s claim limitations requiring “encrypted communications” and “program code.”
The printed matter doctrine a unique and somewhat amorphous concept in patent law that straddles the line between patent eligibility under 35 U.S.C. § 101 and the novelty and non-obviousness requirements of §§ 102 and 103. (more…)
The leading case on copyrightability of AI created works is now pending before the Court of Appeals for the District of Columbia. The case, Thaler v. Perlmutter, No. 23-5233 (D.C. Cir. 2024), centers on Dr. Stephen Thaler’s attempts to register a copyright for an artistic image autonomously generated by his AI system that he has named the “Creativity Machine.” The U.S. Copyright Office refused registration on the basis that the work lacked the required human authorship. Thaler filed suit challenging this determination. The parties have now filed their briefs, along with one law professor amicus brief in support of Thaler.
On February 21, 2024, the Supreme Court heard oral arguments in the case of Warner Chappell Music v. Nealy, which deals with a circuit split over the availability of back-damages in copyright infringement cases. Specifically, the Court is examining whether copyright plaintiffs are barred from recovering damages for acts of infringement that occurred more than three years before filing suit, even if the claims are timely under the “discovery rule.” During the arguments, the parties and justices used metaphors of “bad wine vintages” in questioning the pedigree and value of the discovery rule and damages limitation put before them. [Oral Argument Transcript]
The copyright law has a three year statute of limitations for filing suit that begins when the copyright “claim accrued.” Although “accrual” in copyright cases ordinarily occurs with infringement, the judge-made discovery rule allows copyright claims to accrue when the plaintiff discovers, or reasonably should have discovered, the infringement rather than when the infringing act occurred. All circuit courts accept some form of the discovery rule, but they disagree over a second point — whether damages can reach back beyond the three-year statute of limitations period measured from the date of filing. Although the statute does not make any distinction between a right to file an action and a right to back damages, the distinction stems from ambiguous dicta in the Court’s 2014 Petrella v. MGM decision. Although that case focused on laches, the court suggested general allowance for the discovery rule for copyright claims more than three years from filing, but also wrote that Section 507(b) permits “retrospective relief running only three years back from the date the complaint was filed.” This statement led some courts (including the district court here) to conclude that a copyright holder can sue for wrongful actions that were more than three-years distant, but can only collect damages associated with infringing actions within the three year bubble. In this case, the Eleventh Circuit sided with the plaintiff (reversing the district court) in holding that claims for back damages also follow the discovery rule — permitting potential distant claims.
In the certiorari process, the music publishers asked two broader questions that seemed to potentially challenge the discovery rule itself, but the court granted a re-written narrower question that assumed the discovery rule is in place, but questioned whether it also allows damages beyond the three year mark.
Kannon Shanmugam, representing the music publishers who lost before the Eleventh Circuit, began the argument with a statement that the case “presents a straightforward matter of statutory interpretation” under Section 507(b)’s three-year limitations period. He stated that “a plaintiff can obtain damages for acts of infringement only within three years of filing” and that Petrella supports this position. However, the Justices quickly noted that the question presented “assumes the existence of the discovery rule” and does not revisit whether one should exist at all. This prompted pushback from Shanmugam:
I don’t think that this Court when it rephrased the question presented was accepting any particular version of the discovery rule. . . At most, the rephrased question presented assumes the existence of some version of the discovery rule. It does not take sides on the scope of that rule.
The Justices remained skeptical, with Justice Sotomayor asking “isn’t it artificial for you to raise the most important part of your argument in a footnote” rather than directly asking the Court to revisit the discovery rule. She emphasized that Shanmugam was now trying to make arguments that went beyond the scope of the question presented. Justice Gorsuch echoed this, stating “we’re being asked to decide the scope of something that may or may not exist.”
Respondent’s counsel, Joe Earnhardt, also refused to defend the discovery rule’s validity. He argued the question that was granted certiorari “assumes that there is a discovery rule” and claimed all circuits agree one exists (but disagree as to its scope). Justice Gorsuch then wondered: “Why wouldn’t we just take up that question first” regarding the existence of any discovery rule? But Earnhardt maintained doing so was inappropriate since the issue “wasn’t raised below” and has no circuit split.
Overall, the Justices appear to not be willing to rule expressly on the discovery rule question – whether it exists, even while questioning how to decide a “scope” issue if the rule itself is uncertain. Justices floated the possibility of DIGing the case, meaning dismissing it as improvidently granted (DIG). That had also been suggested by respondent. Justice Alito noted, it makes little sense to “talk about the second [question of scope] without resolving the first [its existence].”
Turning to the core question regarding retrospective damages, Shanmugam argued the time-bar stems directly from the statute of limitations provision that a claim must be brought within three years of accrual. “Our submission is simple. As this Court has said, the standard rule is that ‘accrues’ means at the point when you have a complete and present cause of action.” Plaintiff’s attorney Earnhardt contested this reading. He claimed other parts of the Copyright Act authorize full recovery of damages without any “lookback” restriction, such that imposing one would contradict Congress’s intent. It appears that nothing in the Copyright statute provides any hook for separately treating the damages timing from the general accrual limitation. Justice Sotomayor explained her perspective that “The damages section speaks about damages. The statute of limitations speaks about a time period to file a complaint. You’re automatically tying the two. Tell me how you’re doing it.” Shanmugam argued both that the text demands a three year rule and that the particular damages limitation is derived from the Court’s own prior statements in Petrella. On the other side Earnhardt agreed with the Justices that there is no such connection between the two statutes. This core issue may also help explain the court’s difficulty in sifting through whether they must first answer a question about the discovery rule before analyzing how it applies to damages. Under Shanmugam’s theory, the two are inextricably linked, whereas for Earnheart the statute of limitations is entirely separate and distinct from the copyright damages statute.
Earnhardt insisted Petrella was distinguishable dicta because the plaintiff there “only brought claims for the three-year period” preceding her lawsuit, whereas broader relief would be appropriate if earlier acts were successfully tied to timely claims under the discovery rule. However, Justice Gorsuch noted that Petrella’s language about damages going back “only three years” was seemingly not so limited. Shanmugam argued this was a necessary implication of the prior laches decision — the crux of the decision that laches does not apply to back-damages was the existence of a statute that cuts-off back damages at the three year mark. Shanmugam cautioned that failing limit damages could entrench an overbroad discovery rule that lets plaintiffs reach back indefinitely. During oral arguments, the Court gave little indication of how it might interpret Petrella‘s meaning or perhaps walk-back its language, although Justice Gorsuch sees the case as clearly stating that “you look back three years and no more.”
In the end, the Justices asked few questions about the government’s position that Petrella should not bar damages for timely claims and that the Second Circuit “erred” in Sohm v. Scholastic by holding otherwise based on the Petrella dicta. How the Court maneuvers around Petrella to answer the question remains to be seen when a decision is issued.
The key recurring metaphor during oral arguments was that of “bad wine” in reference to the discovery rule for copyright claims. Justice Gorsuch stated, “we also have a lot of cases in this Court casting doubt on the existence of a discovery rule. We’ve called it wine from a bad vintage or something like that.” See, for example, TRW Inc. v. Andrews, 534 U.S. 19, 37 (2001) (Scalia, J. in dissent writing “The injury-discovery rule applied by the Court of Appeals is bad wine of recent vintage.”); and Rotkiske v. Klemm, 140 S. Ct. 355, 360 (2019) (an “expansive approach to the discovery rule is a bad wine of recent vintage”). Later, Respondent’s counsel argued that the copyright discovery rule is not “bad vintage” because it is inherent to the “accrual” language of the Copyright Act. Of course, the discovery rule is not explicit in the language of the Act, and Gorsuch responded that it “may or may not” be there “some people say that the wine is there.” Sticking to the metaphor, Gorsuch later seemingly referenced the Petrella precedent and subsequent appellate decisions as “not even the old bad wine. It’s something else. It’s a new bad wine.” The “bad wine” metaphor has generally captured skepticism amongst the Justices about a broad discovery rule. After all this back and forth over the vintage, both sides likely needed a smooth evening recovery Merlot. We expect the nine sommeliers to release their opinion by the middle of June 2024.
The Supreme Court is set to hear oral arguments on February 21 in an important copyright case – Warner Chappell Music v. Nealy. The central issue is whether copyright plaintiffs can recover damages for infringing acts that occurred more than three years before filing suit, under the “discovery accrual rule.”
Thus far US copyright and patent tribunals have refused to award rights for AI generated works. China has begun its move in the opposite direction with the recent decision granting rights to an artist who created an image using the popular generative AI system Stable Diffusion. The case of Li v. Liu was decided by the recently Beijing Internet Court. The court decides internet related cases using online tools. Although the court uses AI judges in some cases, this one was decided by humans.
The plaintiff, Li, used Stable Diffusion to generate a photorealistic image of an Asian woman. He shared the work on Xiaohongshu, a Chinese platform similar to Instagram. Later, the defendant Liu, a blogger, published the image in a blog post without Li’s permission, removing his username and platform watermark.
Li sued for copyright infringement. The Beijing Internet Court ultimately ruled in his favor, ordering Liu to apologize publicly, pay ¥500 ($72) in damages, and cover ¥50 ($7) in court fees.
Crucially, the court recognized Li as the legal rights-holder, having contributed the necessary “intellectual input” through choices of models, prompts, parameters, and final image selection. Judges emphasized that copyright law aims to encourage human creativity.
In this case, from the moment the Plaintiff conceived the disputed image to the moment the Plaintiff finally selected the disputed image, it can be seen that the Plaintiff has carried out a certain amount of intellectual input, such as designing the presentation of the character, selecting the prompts, arranging the order of the prompts, setting the relevant parameters, selecting which picture meets their expectations, and so on. . . . Currently, a new generation of generative AI technology is increasingly being used for creation. As long as AI-generated images reflect the original intellectual input of a person, they should be identified as a work and protected under copyright law.
This court ruling signals China’s interest in extending copyright law to better account for AI-generated content and also potentially places the country in the forefront of protecting AI creations. Still, the Beijing Internet Court itself emphasized the approach of judging AI copyright issues based on the specific facts.
Of course, this is one regional lower court decision and is not generally a policy statement of the PRC. The judges here particularly noted that the level of creative expression requires a case-by-case analysis. In an email, Berkeley IP scholar Yuan Hao (郝元) points out, there is still debate within China on this issue. A similar case decided by a court in Shenzhen came to the opposite conclusion, finding AI-generated output eligible for copyright protection as the “intellectual fruits” of the AI developer rather than the AI user. According to Hao, cases like Li v. Liu feature “experimentalist practice” with different local courts exploring the issue before final guidance from the Supreme People’s Court. Hao also disputes depictions of the case as a political move to strategically assist China’s AI industry. While acknowledging that some “political” IP cases may occur, she has not seen evidence of that here. In my view, the low-value of the case (less than $100) also serves as an indication that this was not intended as a landmark decision setting national policy, but rather resolving a minor copyright dispute between two individuals.
The U.S. Copyright Office published a Notice of inquiry (“NOI”) and request for comments, Artificial Intelligence and Copyright, Docket No. 2023-6 on August 30, 2023, calling for comments from interested parties addressing dozens of questions. The Office’s questions focused on a wide range of issues including the copyright implications of the use of in-copyright works as training data, on the feasibility of licensing such uses, the impact on competition and innovation in AI industries depending on how courts resolved training data copyright issues, the copyrightability of AI outputs, whether new laws regulating generative AI were needed, whether AI developers should be obliged to disclose the sources of their training data, and whether AI outputs should be labeled as such.
The Office received roughly 10,000 comments on October 30, 2023. We, who have been writing and teaching about copyright law and how it has responded to challenges posed by new technologies for decades, were among those who submitted comments, see https://www.regulations.gov/comment/COLC-2023-0006-8854.
Substance of the Samuelson, Sprigman, Sag Reply Comments:
We should begin by noting our appreciation for the FTC’s work enforcing both federal antitrust and consumer protection laws and helping to lead policy development in both areas. In our view, the FTC plays a vital role in keeping markets open and honest, and we have long been admirers of the intelligence and energy that the agency brings to that task. More specifically, we recognize the usefulness of examining intellectual property issues through the lenses of competition and consumer protection.
However, in the case of its response to the Copyright Office’s NOI on Artificial Intelligence and Copyright, the FTC has submitted Comments that are unclear and thus open to a variety of interpretations—and possibly to misinterpretations as well. The FTC’s Comments also raise questions about the scope of agency’s authority under Section 5 of the Federal Trade Commission Act, 15 U.S.C. 45, to bring enforcement actions aimed at activities, including those involving the training and use of AI, that might involve copyright infringement—although we would note that the copyright consequences of AI are, as yet, undefined.
We have three principal criticisms of the FTC’s comments:
First, the FTC’s submission is not a model of clarity: indeed, later in these Comments we will focus on a particular sentence from the FTC Comments that is worrisome both for its opacity and for the ways in which it may be interpreted (or misinterpreted) to chill innovation and restrict competition in the markets for AI technologies.
Second, the FTC Comments do not appear to be based on a balanced evidentiary record; rather, the Comments appear largely to reflect views articulated by participants in an Oct. 4, 2023, FTC Roundtable event[1] that featured testimony largely from artists and writers critical of generative AI: 11 of the 12 witnesses appeared to be or to represent individual creators, and one represented open-source software developers who objected to AI training on their code. Not a single witness provided perspectives from technologists who have developed and work with AI agents. Perhaps not surprisingly given the imbalance in the record, the FTC comments do not seem to appreciate the variety of use cases for AI technologies or the broader implications of those technologies for competition policy.
Third, and finally, certain of the FTC’s Comments could, if misunderstood, upset the careful balance that the copyright laws create between private rights to control copyrighted works and public access and use of those works. Upsetting that balance could chill development not only of useful AI technologies, but of a range of new technologies and services that augment consumers’ opportunities to access and use copyrighted works and increase the value of those works to consumers.
In the remainder of these Comments we will focus on a specific sentence from the FTC Comments that illustrates all of these problems. (more…)
Conventional wisdom in copyright law has been to presumptively treat all online content as protected by copyright. This approach has made sense because of the low originality standard for copyright protection. The result then has been that users would focus on whether they have a license; whether the intended use is a “fair use;” and (perhaps) whether the copyright holder is likely to discover the infringement and attempt to enforce their rights.
However, our digital landscape is experiencing a major upheaval with to the proliferation of synthetic works that likely do not meet the criteria for copyright protection. These works primarily fall into two main categories: (1) those lacking the necessary originality for copyright eligibility, such as raw data or certain surveillance camera footage; and (2) those created without direct human authorship, as with content generated by AI systems, often called GenAI.
The existing legal framework requires human creativity and ingenuity as the foundation for originality. Yet, AI systems are generating content at an unprecedented scale, and the volume of synthetic creations has likely already exceeded that of human-originated works. In the upcoming years, we can expect the number of synthetic creations to continue exponential growth. While some of this content may lack originality, a significant portion could easily meet the originality standard established in Feist and similar cases, if not for the absence of a human author.
This emerging copyright-free realm poses both opportunities and challenges. The prospect of freely accessing and building upon an expanding pool of non-copyrighted material without legal ramifications offers a unique freedom. This liberty to transform AI-generated works into new creations could potentially drive innovation and further creation at an accelerated pace. However, a challenge arises in determining to what extent these uncopyrighted synthetic works are discernible from copyrighted counterparts. If my forecast holds — that the majority of works in numerous fields will soon be synthetically generated — it might become prudent to reconsider shifting the burden back onto copyright claimants to explicitly mark their works.
This potential future holds many uncertainties. One question is whether the individuals behind AI-generated works could leverage other legal mechanisms, like contracts, to safeguard their interests. Or, will the courts follow old rules prohibiting the use of contract to limit distribution of otherwise publicly available uncopyrighted content.
The Copyright Act has a seemingly simple three year statute of limitations:
No civil action shall be maintained under the provisions of this title unless it is commenced within three years after the claim accrued.
17 U.S.C. 507(b). It is somewhat surprising then that the Supreme Court has just granted certiorari in a case asking whether the statute “precludes retrospective relief for acts that occurred more than three years before the filing of a lawsuit.” Warner Chappell Music, Inc. v. Nealy, 22-1078 (Supreme Court 2023). At first glance the answer appears to be clearly “no,” but in the case, the Eleventh Circuit sided with the copyright holder by concluding that the rule permits plaintiffs to recover damagers for earlier acts.
The particular dispute focuses on when the “claim accrued” — with the appellate court holding that the three year timeline does not begin until the owner “knows or has reason to know she was injured.” This so called “discovery rule” has been implied into many statutes of limitation, has been rejected for some doctrines. When it decided the important laches case of Petrella v. MGM, the Supreme Court highlighted that “[t]he overwhelming majority of courts use discovery accrual in copyright cases.” 572 U.S. 663 (2014). Still, without directly rejecting the discovery rule Petrella also stated that copyright damages are available “running only three years back from the date the complaint was filed.”
In the Subsequent case of SCA Hygiene Prods. Aktiebolag v. First Quality Baby Prods., LLC, 137 S. Ct. 954 (2017), the Supreme Court provided further input on the discovery rule — although still not deciding directly is applicability.
According to First Quality, § 286 of the Patent Act is different because it “turns only on when the infringer is sued, regardless of when the patentee learned of the infringement.”
This argument misunderstands the way in which statutes of limitations generally work. First Quality says that the accrual of a claim, the event that triggers the running of a statute of limitations, occurs when “a plaintiff knows of a cause of action,” but that is not ordinarily true. As we wrote in Petrella, “[a] claim ordinarily accrues when [a] plaintiff has a complete and present cause of action.” While some claims are subject to a “discovery rule” under which the limitations period begins when the plaintiff discovers or should have discovered the injury giving rise to the claim, that is not a universal feature of statutes of limitations. And in Petrella, we specifically noted that “we have not passed on the question” whether the Copyright Act’s statute of limitations is governed by such a rule.
The SCA Hygiene quote clarifies that statutes of limitations generally start running upon accrual of a claim, not discovery of the injury. The discovery rule that starts the clock upon discovery of the injury is an exception to the norm. The presumption then appears to be that the discovery rule will only be applied to a statute of limitations if there is some particular statutory interpretation reason to do so.
Petitioner’s key argument is that the 11th Circuit’s application of the discovery rule to allow damages beyond the 3-year lookback period conflicts with statements by the Supreme Court in Petrella v. MGM that recovery is limited to three years back from the date the complaint was filed. The responsive briefing pointed to a number of litigation issues, to suggest the case was a bad vehicle for certiorari. On the merits, they also focused on the damages provision, Section 504, that contains no 3-year limitation and also noted that Petrella‘s statements are not controlling because that case only held that laches does not apply as a defense; It did not limit damages to 3 years. In its brief the Chamber of Commerce warns against the discovery rule. The organization represents often-sued businesses and does not want to see a transdoctrinal nationwide discovery rule for federal courts that would extend the statutes of limitations against its members.
Generative Artificial intelligence (GenAI) systems like MidJourney and ChatGPT that can generate creative works have brought a wave of new questions and complexities to copyright law. On the heels of a recent court decision denying registrability of AI created work, the U.S. Copyright Office recently issued a formal notice of inquiry seeking public comments to help analyze AI’s copyright implications and form policy recommendations for both the Office and for Congress. The notice is quite extensive and raises fundamental questions that many have been discussing for several years about copyrightability of AI outputs, use of copyrighted material to train AI systems, infringement liability, labeling AI content, and more. The Copyright Office’s inquiry is an attempt to respond to AI’s rapidly growing impact on creative industries. [Link to the Notice]
The following is a rough overview of three core inquiries that I identified in the notice. It is also easy to just read it yourself by clicking on the notice above.
A core inquiry is whether original works that would ordinarily be copyrightable should be denied unless a human author is identified. Generative AI models produce outputs like text, art, music, and video that appear highly creative and would certainly meet copyright’s originality standard if created by natural people. Further, if human contribution is required, the questions shift to the level of human contribution necessary and procedural requirements to claim and prove human authorship. As the notice states, “Although we believe the law is clear that copyright protection in the United States is limited to works of human authorship, questions remain about where and how to draw the line between human creation and AI-generated content.” Factors could be the relative or absolute level of human input, creative control by the human, or even a word count. With copyright it is helpful to have some bright lines to streamline the process of registration without substantial case-by-case lawyer input for each copyrighted work, but any hard rule might skip over the nuanced. Although the notice focuses on copyrightability, ownership questions will also come into play.
A second important core inquiry focuses on training data that is fundamental to today’s generative AI models. The copyright office seeks input on the legality of training generative models on copyrighted works obtained via the open internet, but without an express license. In particular, the Office seeks information about “the collection and curation of AI datasets, how those datasets are used to train AI models, the sources of materials ingested into training, and whether permission by and/or compensation for copyright owners is or should be required when their works are included.” Presumably different training models could have different copyright implications. In particular, an approach that does not store or actually copy the underlying works would be less likely to be be infringing.
In building the training model, we often have copying of works without license, and so the key inquiry under current law appears to be the extent that fair use applies to protect the AI system generators. In other areas, Congress and the Copyright Office have stepped in with compulsory licensing models, that could possibly work here — a system of providing a few pennies for each web page. Our system also supports approaches to voluntary collective licensing via joint management organizations; perhaps supported by a minimum royalty rate. An issue here is that many of the folks creating training data are doing so secretly and would like to maintain their data and how the model is using the data as trade secret information. That lack of transparency will raise technical challenges and costs for the underlying copyright holders.
A third core area focuses on infringement liability associated with AI-outputs that result in a copy or improper derivative work. Who is liable — the AI system developers, model creators, and/or end users? A traditional approach would allow for joint liability. Again though, the lack of transparency makes things potentially difficult to prove copying, but perhaps availability and likelihood are enough. On this point, notice also asks about the idea of labeling or watermarking AI content as suggested a recent White House / Industry agreement. Although I see this issue as outside of copyright law, the inquiry suggests some penalty for failure to label.
Everyone is floundering a bit in terms of how incorporate generative AI into our world view. I see the Copyright Office AI inquiry as a real attempt to seek creative and potentially transformative solutions. The public is invited to provide input by submitting comments by the October 18, 2023 deadline. There will also be a short response period for reply comments responding to initial submissions that closes on November 15, 2023.
A federal court has dealt a blow to the prospect of granting copyright protections to works created entirely by artificial intelligence systems. In their recent decision, Judge Howell ruled that because AI systems lack human authorship, their output is ineligible for copyright.
The case centers around an image generated by an AI system called the Creativity Machine, which the system’s owner Stephen Thaler attempted to register with the U.S. Copyright Office. Dr. Thaler is the same individual who unsuccessfully attempted to protect and invention created by a separate machine that he had titled DABUS.
In this case, Thaler listed the Creativity Machine as the author and designated the work as created autonomously by the AI, with ownership transferring to himself as the owner and creator of the machine. The Copyright Office denied the registration application on the grounds that copyright law requires human authorship. Thaler challenged the rejection in court, but Judge Howell upheld the Copyright Office’s decision.
In her ruling, Howell emphasized that human authorship has underpinned copyright law since its Constitutional inception, which enables Congress to grant exclusive rights to “authors.” While copyright has adapted over time to cover new technologies like photography and film, Howell wrote, it has never extended so far as to protect works created without any human involvement. The court does not discuss in any detail the work made for hire doctrine that does permit non-human authorship. However, even in that situation a human underlies the creation.
Judge Howell also noted that machine learning systems like the Creativity Machine do not require copyright incentives, since they operate algorithmically rather than responding to legal rights and protections. The opinion concludes unambiguously that “in the absence of any human involvement in the creation of the work, the clear and straightforward answer is the one given by the [Copyright Office]: No.”
While this ruling provides clarity for now, the accelerating development of AI systems means the issue is unlikely to disappear. As AI grows more advanced and autonomous in generating creative works, questions around copyright protections will likely resurface. Lawmakers may eventually need to reconsider whether AI-created works should be eligible for copyright, especially if human direction becomes minimal or nonexistent. With the current copyright regime requiring human authorship, however, true machine-created art remains in the public domain.
While this case dealt with a work claimed to be created entirely autonomously by an AI system, a more common scenario today sees individuals using AI tools in tandem with their own creativity and direction. In these situations where both human and machine contribute, questions arise around how much and what kind of human input is necessary to meet the originality and authorship requirements for copyright protection. If an artist uses a generative AI tool to create a work but provides extensive creative guidance and selection, there would seem to be a strong argument for human authorship. However, as AI grows more advanced and autonomous, that human contribution may become limited to little more than prompting the system. In these cases, courts will need to draw difficult lines around what amounts to human authorship. If AI progress continues apace, the fundamental question of whether true machine-created works should be eligible for copyright will require legislative attention.
A few days ago I wrote an essay about the ongoing economic war between the leaders of the US and those of China, with a focus on the impact US attorneys representing Chinese clients. For the article, I used an AI tool to create some art for my publication. The image above shows the resulting four images. I chose the bottom right and had the AI enhance the image for publication by pushing a single button. I then did a cut and paste before publishing.
Although I didn’t contribute any of the expression seen in the work, after these actions and then publishing it, I actually feel some gut level of ownership. And, that I might be upset if someone else used _my_ image without permission.
My questions for you:
Do I deserve to “own” this image?
Should I be able to claim authorship?
Should I disclose my AI creation methodology when I use the image in published works?
The copyright lawsuit between the data-software company SAS Institute and its scrappy copycat World Programming has been interesting to follow over the past several years, and the Federal Circuit has now issued a controversial opinion in the case. SAS Inst. v. World Programming Ltd., — F.4th — (Fed. Cir. 2023). The majority opinion authored by Judge Reyna and joined by Judge Wallach affirmed the lower court ruling that SAS failed to establish copyrightability of its claimed program elements. Writing in dissent, Judge Newman argued that the majority’s rejection of copyrightability represents a “far-reaching change” not supported by either precedent or good policy. I called this outcome controversial. The outcome would also be controversial had Judge Newman’s position prevailed.
The case is properly seen as an extension of the Supreme Court’s decision in Google LLC v. Oracle America, Inc., 141 S. Ct. 1183 (2021). In that case, the Court found that Google’s use of Java API naming conventions in its Android operating system was fair use under copyright law. Because its fair use decision decided the case, the court did not rule separately on whether the API was even copyrightable in the first place. In SAS v. WPL, the Federal Circuit squarely addressed the copyrightability question.
To be clear, computer software can still be copyrightable. But, parties asserting protection will need to do a much better job of showing how their creative authorial input survives the “abstraction-filtration-comparison test,” which the Federal Circuit applied in its decision.
Copyright law’s abstraction-filtration-comparison (AFC) test is used to determine whether a particular work is entitled to copyright protection. The AFC test involves breaking down a work into its constituent parts, abstracting the unprotectable elements, filtering out any remaining unoriginal or unprotectable elements, and then comparing the remaining protectable elements to the allegedly infringing work. The AFC test has been previously adopted by the Second, Fifth, and 10th Circuits.
Here, the court did not delve into the comparison step — and instead simply held that there was nothing left to infringe after abstraction & filtration.
The decision is also substantially procedural. The district court held a copyrightability hearing and followed a burden shifting procedure created by the 11th Circuit in Compulife Software Inc. v. Newman, 959 F.3d 1288 (11th Cir. 2020). In particular, the court first assumed that the work was copyrightable based upon the registration documents. It then allowed the defense to present its filtration argument to show a lack of copyrightability. If that evidence is sufficient (as it was here), the burden then shifts back to the copyright holder to rebut — and “to establish precisely which parts of its asserted work are, in fact, protectable.” The difficulty for SAS is that it offered no rebuttal and instead “refused to engage in the filtration step and chose instead to simply argue that the SAS System was ‘creative.'” Slip Op. SAS presented an expert witness on copyrightability, but the district court found it extremely unreliable and thus excluded the testimony. (The expert had not seen anything to filter out — even clearly unprotectable elements).
The majority walked through each of these issues and ultimately affirmed on all grounds.
= = = =
Some background: SAS makes data analysis software. A key feature of the SAS product is that folks can write programs using SAS syntax in order to get certain results. Thus there are a number of data jockeys who are experts in SAS code. WPL is a UK based software company who obtained several copies of SAS statistical software and made their own clone version by rewriting the code and by relying upon an early version of SAS that is not protected by copyright. The WPL version allows folks to use SAS language to get the same results — but at a much lower price. When I sa “same results” — the clone pretty much identically copied output styles so that a chart made with WPL looks basically identical to a chart made in SAS using the same code.
SAS sued in E.D.Tex for copyright infringement. Judge Gilstrap dismissed the copyright claims — holding that the software was unprotectable. Copyright infringement appeals are ordinarily not heard by the Federal Circuit, but in a case of what appears to be appellate-forum shopping, SAS had also included patent infringement allegations that they eventually stopped pursuing. Under the rules of procedure, if patent claims were raised in the case at some point, then the appeal heads to the Federal Circuit.
The copyright case is not about copying code. It appears rather to be about copying the input syntax format used by individuals to input their programs and the output design styles for outputting data in some particular style. In the filtration analysis, WPL provided a host of evidence to show that these features should be “filtered out” of the SAS copyrights.
WPL established that an earlier version of the SAS System, “SAS 76,” was in the public domain.
WPL showed that many Input Formats and Output Designs in the current SAS System are identical or nearly identical to those in SAS 76 and should be filtered.
WPL demonstrated that the SAS Language should be filtered because it is open and free for public use.
WPL’s expert identified various allegedly copied materials that contained unprotectable elements such as open-source, factual, data, mathematical, statistical, process, system, method, and well-known and conventional display elements.
Bringing these together the Federal Circuit concluded that the defense had presented sufficient evidence to show uncopyrightability and that the district court was justified in requiring SAS to directly and particularly rebut the evidence rather than simply allowing a trial on the copyright as a whole.
The district court was correct to exercise its authority and require SAS to articulate a legally viable theory on which it expected to base its copyright infringement claims. Conversely, it would be improper for a district court to permit a matter to proceed to trial on the basis of vague and unidentified theories.
Slip Op.
Writing in dissent, Judge Newman argued that Fifth Circuit law protects this sort of computer software architecture even from non-literal copying. The key citation is likely to a the Fifth Circuit’s 1994 Engineering Dynamics case:
Most courts confronted with the issue have determined that copyright protection extends not only to the literal elements of a program, i.e., its source code and object code, but also to its “nonliteral” elements, such as the program architecture, “structure, sequence and organization,” operational modules, and computer-user interface.
Eng’g Dynamics, Inc. v. Structural Software, Inc., 26 F.3d 1335 (5th Cir. 1994). Judge Newman noted that “computer programs” are expressly protected within the Copyright Act
Copyright protection subsists . . . in original works of authorship . . . including . . . (5) computer programs.
17 U.S.C. 102. As the Nimmer treatise explains, this 1980 amendment to the laws “dispels any lingering doubts as to the copyrightability of computer programs. It is
therefore now firmly established that computer programs qualify as work of authorship in the form of literary works, subject to full copyright protection.”
1 NIMMER ON COPYRIGHT § 2A.10(B) (2022 ed.).
Here, Judge Newman particularly noted that the collection of the various input functions and output designs is easily copyrightable. And, this is the same analysis done by the Federal Circuit in its original Oracle v. Google decision.
Judge Newman also concluded that the district court improperly shifted the burden of proof to the copyright holder.
Rights to use a person’s Name-Image-and-Likeness or NIL generally fall within two categories of intellectual property: rights of publicity and rights of privacy.
Rights of publicity protect commercial value that a person has developed in becoming a celebrity. Rights of publicity can also protect against resulting false impressions created by a seeming endorsement.
Rights of privacy can also protect against misappropriation of NIL — typically under the guise of an invasion-of-privacy claim.
The pending case of Gibson v. RPS Holdings LLC, 5:21-cv-00416 (E.D.N.C. 2023) involves a set of thirteen different professional models whose images were allegedly used without their permission to advertise for the Capital Cabaret, a strip club halfway between Raleigh & Durham, North Carolina. None of the models have any history with the club. Apparently, the club obtained photos of the models; and then edited them into advertisements for pubilcation on Facebook, Instagram and other online media.
One difficulty for professional models such as the plaintiffs here is that copyright the photographic images is typically held by the photographers, not the models. But, modeling agreements also regularly include limited licenses regarding how (and for how long) the images will be used.
In Gibson, the plaintiffs have sued under both the Lanham Act and N.C. state law:
Section 43 of the Lanham Act, 15 U.S.C. § 1125(a)(1) for misrepresentation of sponsorship (False Advertising + False Association).
N.C. common law right of privacy – misappropriation.
N.C. Unfair & Deceptive Trade Practices.
N.C. Defamation.
Etc.
The case is moving forward and is at the close of discovery (March 31, 2023) with summary judgment motions due at the end of April.
Most recently, plaintiffs filed a motion to quash a subpoena seeking information on exactly how the attorneys “discovered, came across, learned of, and/or otherwise became aware of” the advertisements. The motion argues that this information is protected by attorney client privilege as well as work product. The defense argues that this information is important for its laches defenses.
I mentioned the copyright difficulty above. A second difficulty for the models in this case is that most rights-of-rights of publicity doctrines require focus on celebrity; and ask whether that person’s NIL is identifiable to the consuming public. I have to admit that I don’t really know anything about the modeling industry, but I don’t believe think that these models are household celebrities. The ads from the strip club were not intended to indicate that these particular models support the club; rather, the intent was to provide a provocative image.
Of course the misappropriation of a completely anonymous face could not form the basis for a false endorsement claim, because consumers would not infer that an unknown model was ‘endorsing’ a product as opposed to lending her image to a company for a fee.
Bondar v. LASplash Cosmetics, No. 12-cv-1417, 2012 WL 6150859, at *7 (S.D.N.Y. Dec. 11, 2012) quoted in Electra v. 59 Murray Enterprises, Inc., 987 F.3d 233, 258 (2d Cir. 2021), cert. denied, 211 L. Ed. 2d 352 (Nov. 22, 2021). The Carmen Electra case was also a suit by models against strip clubs with the models represented by the same law firm. (The Casas Law Firm). In their denied petition for certiorari, the plaintiffs asked:
Must an individual prove they have a commercial interest in their identity, or must a person prove they are recognizable, publicly prominent, or a celebrity, to bring and sustain a claim under 15 U.S.C. § 1125(a)?
Petition. The Electra case was decided by the 2nd Circuit. However, other circuits have rejected the celebrity requirement as improperly grafted-onto the statute.
In a prior post, I mentioned that the Copyright Office had canceled the registration for “Zarya of the Dawn,” a book purportedly created mainly by AI. That was in error apparently generated by the Office’s new Copyright Public Records System. The attorney for the human author – Kristina Kashtanova – contacted me to point out the error and you can see that the errors have been corrected. The copyright is currently registered.
That said, the Copyright Office has initiated a cancellation process for the work. In a responsive filing, her attorney Van Lindberg walks through her creation and makes the case that all aspects of the book, even computer generated images, are copyrightable.
In addition to the copyrightability of the Work as a whole, each individual picture is itself the result of a creative process that yields a copyrightable work. Kashtanova could extract any single image from the Work and submit it to the Office and correctly assert her authorship of that image.
Unlike the “autonomously generated” picture known as “A Recent Entrance to Paradise,” all the images in the Work were designed by Kashtanova. The visual structure of each image, the selection of the poses and points of view, and the juxtaposition of the various visual elements within each picture were consciously chosen. These creative selections are similar to a photographer’s selection of a subject, a time of day, and the angle and framing of an image. In this aspect, Kashtanova’s process in using the Midjourney tool to create the images in the Work was essentially similar to the artistic process of photographers – and, as detailed below, was more intensive and creative than the effort that goes into many photographs. Even a photographer’s most basic selection process has been found sufficient to make an image copyrightable. The same reasoning and result should apply to the images in Kashtanova’s Work.
I previously wrote about the copyright preemption case of ML Genius v. Google that is pending before the Supreme Court. The court took a step toward granting certiorari with its request for the Government to file an amicus brief (CVSG). Genius doesn’t hold copyright to the song lyrics that it has on its website, but it does (arguably) require users of its site to contractually agree not to copy and use those lyrics for commercial purposes. Google scraped the site and (apparently) is using the lyrics for its own commercial gain. The appellate court sided with Google, holding that the Copyright Act preempts a breach-of-contract claim because it is “equivalent to … exclusive rights within the general scope of copyright,” 17 U.S.C. § 301(a). Other circuits have held that similar contracts are not preempted by the Copyright Act.
Question presented: Does the Copyright Act’s preemption clause allow a business to invoke traditional state-law contract remedies to enforce a promise not to copy and use its content?
Petition. The SG’s office already has three other CVSG briefs that it should now be drafting in patent cases:
Tropp v. Travel Sentry (eligibility)
Interactive Wearables v. Polar Electro (eligibility)
Teva v. GSK (FDA/Patent skinny label infringement)
Certiorari tends to be much more likely if the SG supports the case. That said, the SG did not support the one patent case granted certiorari this term. Amgen v. Sanofi (enablement).
by Dennis Crouch
by Dennis Crouch
