AI Generated Art and Copyright Infringement

Artificial Intelligence is coming to art near you! And with AI’s entrance into the field through systems like Stable Diffusion, DALL-E, and Midjourney, copyright infringement lawsuits are the next wave as we try to sort out the technology and the rights of persons affected by the technology.

The use of AI in generating art online has become increasingly popular. Many artists and designers are using AI tools to create new and unique pieces of art, from digital paintings to animations and more. The problem is how these AI systems learn how to respond to text prompts from system users.

One approach to training AI systems is to use images scraped from the internet – regardless of the copyright status of the image – and to use this library of images to inform the algorithm on how to interpret a text prompt from a system user. So, to simplify, if a system user enters a text message such as “please make an image of a cow standing in a green field on a rolling hill,” the AI algorithm would need to have reference material of what a cow, green field, and rolling hill looks like. The reference images that are indexed in relation to these words and phrases help the AI to learn how to respond to the text prompt and generate a new image.

Figure 1 – DALL-E generated image of a “cow standing in a green field on a rolling hill”

The problem comes in with the images that are used by the AI in its original training, as some aspect of these images must be re-used by the AI to respond to a user text prompt. If the copyright owner of the source image has not dedicated the image to the public domain, or otherwise licensed the work openly, an AI that works based on combinations of these protected works may infringe a copyright right (such as the exclusive right of duplication, derivative work, etc.) under section 106 of the Copyright Act.

And several lawsuits along these lines have been filed, challenging the use by AI companies of these original source images. For example, Anderson v. Stability AI is a putative class action lawsuit brought by artists alleging infringement against Stability and several other AI image generation businesses. Getty Images has also filed a similar type of infringement suit. In Getty’s case, the images it owns have a proprietary watermark that permits the company to track unlicensed use of works, and alleges that “millions” of its images have been used by companies like Stability AI without license or consideration.

This leads to the unanswered question of whether the use of these protected works is infringing, and if it is, whether the use could be a defensible fair use under section 107. A key question for the source works harvested from internet sources is whether they are sufficiently original to be protected. The bar of originality through court decisions here is relatively low. Feist v. Rural Tele., 499 U.S. 340 (1991). Absent slavish copying of an original, most photographs will have some originality, at least as to photographer choices about composition and lighting, though the copyright interest in photographs of common objects and scenery must be relatively narrow, which is another way of saying that an infringing work must be nearly identical for liability to accrue. However, there really is no fixed formula of originality of all photographs and the degree of similarity for a subsequent work to be infringing. Such issues require a judgment on a case-by-case basis. For example, are Warhol prints based on a photograph of the late musician, Prince, sufficiently different to not be a copy of the protectible aspects of the photograph? The Second Circuit judged them to not be sufficiently different and to also not be sufficiently transformative to be protected under fair use. Even a casual reading of the literature on fair use would find the wide range of decisions and the absence of simple rules in this area. Andy Warhol Foundation for the Visual Arts v. Goldsmith, 992 F.3d 99 (2d Cir. 2021) (oral arguments heard at US Supreme Court Oct. 12, 2022).

Another problem of duplication involves the doctrine of scenes a faire, which essentially is the idea that genres or subject matters of works will necessarily share some common ground. Feist, 499 U.S. at 340. All works of penguins will share commonality in the “penguinness” of the birds pictured, therefore my photograph of penguins cannot prevent all others from taking their own photo of that bird at the zoo or in the wild. Teaching AI what image relates to the word “penguin” necessarily means showing the AI many penguin images for it to construct some generalized model of “penguin.” Therefore, copyright holders would not be able to prevent AI from generating images of penguins, even images that are based on images made by others, as all pictures of penguins necessarily include aspects of the bird’s form, markings, and size that make it a penguin rather than other birds or animals.

Another unresolved issue with this situation is the idea of sampling. The Ninth Circuit discussed several sampling cases – where a small portion of a work is used by another such that the average observer would not notice the appropriation – and explained that such sampling is generally not actionable infringement. Bell v. Wilmott Storage, 12 F.4th 1065 (9th Cir. 2021) (discussing Newton v. Diamond, 388 F.3d 1189 (9th Cir. 2004). A court could view the AI use of copyright materials of other as a mere sampling of that protected material, and therefore not actionable if the AI were to stitch together hundreds of pieces of digital files to render a response to a user-supplied text prompt.

A final fair use issue has to do with Google’s admitted copying of millions of paper books and indexing of those books into Google’s index of these works, without permission or license from the copyright holders. The Second Circuit ultimately affirmed the finding of fair use in this case, partly on the grounds that works that were still within copyright, while indexed in the database, only appear to searchers in “snippet” form that displays a limited portion of the relevant book page with the search result. Author’s Guild v. Google, Inc., 804 F.3d 202 (2d Cir. 2015). In addition, many courts will examine whether the use of the plaintiff’s work is “transformative,” though this determination often can be viewed as subjective. For example, the second circuit did not find Andy Warhol’s use of Goldsmith’s photograph of Prince as transformative, but the same circuit has found numerous other appropriationist artist’s use of another’s work as transformative, such as in Cariou v. Prince, but not in Rogers v. Koons, yet confoundingly Koons’ use of a different photograph in Blanch v. Koons was sufficiently transformative. Andy Warhol, supra; Cariou v. Prince, 714 F.3d 694 (2d Cir. 2013); Rogers v. Koons, 960 F.2d 301 (1992); Blanch v. Koons, 467 F.3d 244 (2d Cir. 2006). Perhaps a court could view AI’s use of photographs scraped from the internet as sufficiently transformative, as AI generates a “new” image based on the user text prompt from relevant reference material.

While the source materials are photographs that are published on the internet, a court would most likely view them as within the core protection of the Copyright Act, so this factor would likely favor photographers whose works have been used by AI to generate images. Cariou, 714 F.3d at 710.

The third factor, the amount and substantiality of the protected work used by AI to generate the subsequent image to the AI user, is less certain as to which party is favored. Id. We do not really know how much of the scraped images the AI actually uses in the finished product it displays to users of these various systems, or whether the AI has used more than is necessary to accomplish its transformative purpose.

Finally, courts would have to examine whether the AI’s use “takes away from”usurps” the existing market for the original photographs. Id. at 709. The market for AI-generated art is relatively new. DALL-E does have a paid subscription level for image generation beyond a minimum number of free images; whereas Midjourney is a subscription-based system that starts at $96 per year.  What remains to be determined is whether these paid systems reduce the market value of works in Getty Images’ database. For example, are AI users merely generating images with the AI system that they would otherwise just license from Getty Images? While this may seem unlikely from a cursory use of the AI systems, what evidence offered in court of market impact may support either party’s contentions here. However, if the AI systems are generating income, the door is open for these systems to enter into some licensing system with the works they have used to go “legitimate.”

Stay tuned for what’s next and further market disruption by AI systems available and yet to come to market.

Arbitration Clauses in Maryland

Arbitration Clauses in Contracts

An arbitration clause is a provision in a contract that requires disputes to be resolved through arbitration rather than in court. The general purpose of arbitration clauses is to provide a more efficient and cost-effective way for parties to resolve disputes. There is a strong federal and Maryland public policy that favors arbitration where parties have agreed in advance to resolve disputes in that manner, as this reduces the load on the civil court system, and also respects the voluntary, private choices of persons and businesses.

Benefits for companies can include:

  • Lower costs than litigating in court, as arbitration proceedings generally do not involve appeals of decisions made by the arbitrator.
  • More control over the process and the outcome.
  • Confidentiality of the proceedings and the outcome. Unlike civil court proceedings, where the public has a right to access non-privileged records and information about court proceedings are available through online resources such as the Maryland Judiciary Case Search, arbitrations are confidential by default.
  • The possibility of a more favorable outcome, as arbitrations preclude jury trials that may be more favorable to consumers, and also preclude class action lawsuits against businesses. Instead, aggrieved consumers must litigate individually in an arbitration proceeding.

Benefits for consumers can include:

  • Potentially lower costs than litigating in court, though consumers need to take into account the cost of the arbitrator, which generally must be shared with the other party and must be paid in advance to file an arbitration.
  • A faster resolution than in court. Arbitrations can be resolved on average in less than a year, where actions in state courts like the Maryland circuit court may take years (https://go.adr.org/impactsofdelay.html)

However, there are also costs and drawbacks to consider:

  • Consumers may have less protection under arbitration than in court, especially if they are not familiar with the process or do not have the resources to hire an attorney. Generally, arbitrations are governed by different rules of procedure than civil cases, which in some cases may limit rights to obtain discovery, or limit other litigation steps.
  • Consumers may be limited in their ability to participate in class action lawsuits, which can provide more leverage in disputes, and also permit pooling of small claims for resolution making the class action more cost-effective for consumers in some cases.
  • Companies may use arbitration clauses to limit consumer rights and protections.

The Maryland federal district court recently decided a case involving a broad, mandatory arbitration clause in an employment agreement between a car dealership and a former manager. Brown v. Brown’s Md. Motors, Inc. (U.S.D.C. June 10, 2022). In that case, the plaintiff had entered into an employment agreement that contained a broad arbitration clause, and was subsequently terminated from his employment. The plaintiff sought to have his day in court against his employer, but his employer filed a motion to compel arbitration. In response, the court analyzed plaintiff’s claims as to the enforceability of the arbitration clause, applying Maryland contract law to the dispute. Such defenses can include: (a) claims that the agreement was not entered into or was otherwise defective, (b) applicable contract defenses like unconscionability, or (c) in this case, claims that arbitration would prevent the plaintiff from vindicating important federal rights. Ultimately, the court dismissed these claims and ordered arbitration.

If you are considering an arbitration agreement, schedule a consultation on Zoom to review the details before you enter into the agreement.

Is AI Coming for Your Law Job?

Artificial Intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. AI systems can be trained to perform tasks such as recognizing speech, understanding natural language, and making decisions.

Legal research is a primary target for an AI intervention, as current legal research tools are fragmented and require advanced knowledge by the human user to identify relevant authority. These tools can also be time consuming, as the human user often still must sift through search results to evaluate if the case, statute, or other authority is most relevant to the assigned research task.

Google had made some strides in this area of research when it published Google Scholar. This research tool incorporated publicly available cases across all US jurisdictions and indexed them. This research tool could be more effective in taking natural language search phrases and finding relevant results, though it lacks other research tools to validate that cases or statutes remain in force.

Another key issue is that different research databases may return different results, and therefore a careful researcher may have to consult multiple databases, including official paper-based reporters, to get a complete picture on the applicable legal authority.

AI has been used in the practice of law in several ways, including:

  1. Legal research: AI-powered legal research tools can analyze large amounts of legal text and help lawyers quickly find relevant case law and statutes. For example, LexisNexis has a “brief analysis” product within Lexis+ that utilizes AI to quickly summarize cases for legal researchers. Brief Analysis in Lexis+
  2. Contract review: AI-powered contract review tools can assist lawyers in analyzing and summarizing the key terms and provisions of contracts. A variety of vendors offer solutions in this area, such as Foley & Lardner LLP, LinkSquares, and Klarity.
  3. Litigation Support & eDiscovery: AI-powered predictive coding tools can help lawyers identify relevant documents in large sets of data, such as during discovery in litigation, though debate continues as to whether AI lives up to the marketing hype surrounding litigation support. Law.Com (2022)
  4. Chatbots: AI-powered chatbots can assist lawyers in answering frequently asked legal questions and guiding clients through legal procedures.

In the court system, AI has been used for tasks such as:

  1. Sentence classification: AI-powered tools can assist judges in determining appropriate sentences for defendants based on factors such as criminal history and the nature of the crime, though such use is not without concerns about bias and due process. Hillman, Noel (2019)
  2. Predictive policing: AI-powered predictive policing tools can assist law enforcement in identifying areas where crimes are more likely to occur and allocating resources accordingly, though such tools may also lead to claims of racial and ethnic bias. Verma, Pranshu (2022) Washington Post

Overall, AI is being used increasingly in the legal system to analyze data and make predictions, but it is important to note that the implementation and use of AI in the legal system is still in its infancy, and there are concerns regarding bias and accountability. AI holds the promise of improving knowledge worker productivity, for example, to aid a human knowledge worker in more quickly identifying relevant authority and summarizing it for the human knowledge worker. Important limitations on AI remain, however. For example:

  1. Complex legal reasoning: The legal system is complex and requires a deep understanding of the law and its application. AI systems may struggle to fully grasp the nuances of legal reasoning, making human lawyers more effective in this area.
  2. Communication and negotiation: The legal system requires human interaction and communication, as well as the ability to negotiate and come to agreements. AI systems may not be able to fully replicate the emotional intelligence and interpersonal skills of human lawyers.
  3. Ethical considerations: The legal system involves many ethical considerations, such as the protection of individual rights, which may be difficult for AI systems to fully comprehend. More generally, concerns exist about the ethical use and implementation of AI (Garibay, Ozlem, et al. (2023))
  4. Creativity: There are cases where creative thinking is required to find solutions, AI systems may not have the ability to think creatively.
  5. Understanding of social, cultural and economic context: In the legal industry, understanding the broader context of laws, regulations, societies and cultures is important in order to make informed decisions.

However, with the rise of ChatGPT and the substantial reported investment in this platform by companies like Microsoft (Browne, Ryan (2023) CNBC) suggests that 2023 will likely be a year to watch for AI in a job near you!

The FTC and Non-Competition Agreements

Earlier this month, the Federal Trade Commission proposed a rule that would ban the use of non-competition clauses for employees, on the grounds that such clauses constitute an unfair method of competition, in violation of section 5 of the Federal Trade Commission Act. FTC Press Release

Non-competition agreements, also known as non-compete clauses, are agreements in which an individual agrees not to work for a competitor for a certain period of time after leaving a job. The enforceability of non-competition agreements varies by state in the United States. Some states, such as California, generally do not enforce non-competition agreements because they are viewed as unreasonable restraints on trade. Cal. Att’y General Other states such as Maryland will enforce “reasonable” non-competition clauses depending on the geographic extent and duration of such clauses. Holloway v. Faw, Casson & Co., 319 Md. 324 (1990).  The ostensible purpose in enforcing such clauses has been to protect an interest in the goodwill of the employer that might be invaded by a departing employee that turns to a competitor with a book of clients obtained through work at the former employer. Fowler v. Printers II, 89 Md. App. 448 (1991). The proposed FTC rule acknowledges that this sort of promise to not solicit customers, however, is not a violation of section 5 of the FTCA. Instead, the FTC rule seems to be aimed at making unenforceable those clauses that prevent a worker from seeking employment at a different business or starting their own, competitive business with a former employer. § 910.1(b)(1).

The FTC’s comments estimate that approximately thirty million workers or about twenty percent of the nation’s workforce is subject to a non-competition clause. Studies cited by the FTC support its conclusion that preventing the enforcement of non-competition clauses may increase wages for workers. For example, the FTC discusses research by Matthew S. Johnson, et al., that concludes that an increase in enforcement of non-competition clauses tends to reduce workers’ earnings and exacerbates gaps in wages based on gender and race. (Johnson, 2021)

In particular, the study authors discuss that wages decline by 3.5% when comparing worker wages in a state that is less likely to enforce a non-compete clause with one that is very likely to. The authors also estimate that if all non-compete clauses were unenforceable nationally, average earnings among all workers would rise between 3.3% and 13.9% (Johnson, 2021, pps 18-19).

Undoubtedly, the FTC rule will be subject to court challenges as to its enforceability under FTCA in the coming months. Concerned about a non-competition clause in an agreement? Reach out to schedule a review of your agreement.

Fair Use in the Copyright Act

Section 107 of the Copyright Act provides for a “fair use” defense to claims of copyright infringement. The statute generally requires the application of a four-factor test to determine whether the defendant’s use of the plaintiff’s work, while infringing, is still permissible under the Act. The problem with this statute is that its application is not self-evident in numerous cases involving contemporary art.

The first element of first use, the purpose of the defendant’s use of the plaintiff’s work, involves a factual analysis of why the defendant did what she did with the plaintiff’s work. On the one hand, a defendant merely trying to profit from a plaintiff’s original work is probably not a fair use purpose, while a defendant that “transforms” the original work into a new expression may be a fair use purpose.

Recently, the Second Circuit had to address this issue in the case, Andy Warhol Foundation for the Visual Arts v. Goldsmith, 11 F.4th 26 (2nd. 2021) (cert. granted by USC Mar. 28, 2022). At issue in that case was a photograph taken by the Goldsmith of the late, well-known musician, Prince, which was subsequently used by the late Andy Warhol in a series of prints based on the photograph. On discovery of the unlicensed use, Goldsmith notified Andy Warhol Foundation for the Visual Arts (“AWF”) of the apparent infringement, and AWF subsequently filed a lawsuit for a declaration that the Warhol works were not infringing as a matter of law.

Appropriation artists are no stranger to copyright infringement lawsuits in the second circuit. Jeff Koons was subject to two different cases involving his use of photographs in his works. In Rogers v. Koons, 960 F.2d 301 (2nd cir. 1992), the Rogers Court found that Koons’ use of the photo was not a fair use as the use was not sufficiently transformative. However, in a later case, Blanch v. Koons, 467 F.3d 244 (2006), the Court found that Koons’ use of a different photograph was a fair use. Both, of course, were appropriations of another’s work, but the distinction of which was fair under the copyright statute turned on the Blanch court’s finding that the use was sufficiently transformative.

Another appropriation artist, Richard Prince (no relation to the late musician, Prince), was subject to a lawsuit involving copyright infringement by yet another photographer, Patrick Cariou. Cariou v. Prince, 714 F.3d 694 (2nd cir. 2013) (cert. denied). There, the Cariou court found that most of Prince’s paintings and collages were fair uses of Cariou’s works as a matter of law, and remanded the case to the district court to evaluate whether the remaining five were also fair use using the standards laid out in the appellate court’s decision.

In Goldsmith, however, the Second Circuit decided that Andy Warhol’s use was not a fair use of Goldsmith’s unpublished photograph of the late Prince. The Court found that Warhol’s removal of certain elements from Goldsmith’s photograph, and embellishing them with loud and unnatural colors was simply not transformative. While the Court is careful to insist that judges should not be transformed into art critics to decide fair use claims, the Court’s attempt to cabin in fair use to some objective standard for works may just be the Court’s rejection of “appropriationism.”