Category: Law & Policy

Public Knowledge Post on AI & Fair Use Misses the Mark

fair use

Patrick Gallaher at Public Knowledge recently posted an article about AI training with protected works, proposing to distinguish between piracy and fair use. Not to begin on a pedantic note, but the article is subtitled “Words Matter” because it claims that piracy is a provocative, non-legal term, so I have to respond by saying this is wrong. Although we think of “piracy” today as enterprises like The Pirate Bay, courts have often used the term “piracy” to mean “copyright infringement.” For instance, the seminal fair use case Folsom v. March (1841) uses the word thirteen times as in this quote:

“….it is as clear, that if he thus cites the most important parts of the work, with a view, not to criticise, but to supersede the use of the original work, and substitute the review for it, such a use will be deemed in law a piracy.”

So, Gallaher is making a semantic fuss over nothing. If a contemporary court holds that AI training with protected works is copyright infringement, then this conduct may both legally and colloquially be called piracy.

As to the substance of the post, Gallaher asserts that AI training is inherently fair use, which is too broad a claim. The fair use doctrine defies generalization, and the facts in one case involving a particular AI and one type of work may have limited influence on the result of a case involving a different AI and different type of work. Or to put that another way, the incomplete fair use inquiry conducted in Bartz v. Anthropic, involving a class of literary works, likely predicts almost nothing about the eventual outcome in UMG et al. v. Udio or Disney et al. v. Midjourney, involving sound recordings and visual works respectively.

Gallaher states that AI training is transformative under fair use factor one (the purpose of the use). Indeed many articles of this nature rely on the assumption that this finding should be obvious and should carry the weight of the fair use analysis. “Copying for training is transformative: it uses the works for a fundamentally different purpose from the original, much like indexing websites for search engines or scanning books for text analysis,” he writes. And that’s all he writes about one of the most vexing doctrines in fair use weighing the most challenging technology ever confronted by copyright law.

Of course, even in one sentence, Gallaher manages to hide (or expose) the distinction that the purpose of many GAI products is to produce works without authors. This fact is highly distinguishable from the two analogies he cites and, as the courts will surely recognize, presents a novel challenge to the constitutional intent of copyright law. This is a consistent fallacy with every article of this nature—claiming that AI is the most revolutionary tech in history, but despite this novelty, we have ample case law to conclude that training is fair use.

Perhaps the courts will not wholly agree with my view that a purpose which does not serve the goals of copyright cannot favor fair use, but in Kadrey v. Meta, Judge Chhabria stated, “Courts can’t stick their heads in the sand to an obvious way that a new technology might severely harm the incentive to create, just because the issue has not come up before.”

Although that sentence prefaces a consideration of market dilution under factor four, the words “harm the incentive to create” allude directly to copyright’s core purpose and, so, implicates the purpose of the GAI to “create” in lieu of authors. And that goes to the question of transformativeness. So, no, it is not enough to say that a use which serves a different purpose is per se transformative, especially when that different purpose is to do exactly what creators do and, in the process, moot the utility of copyright law.

Notably, Gallaher masks the substitutional purpose of GAI by referring to it in general as technology that serves a “public good” and which provides “broad benefits.” The plain fact, though, is that we do not know this to be true. Simply because a product is new, being widely adopted, and/or has investors chomping at the bit is not evidence that its purpose is categorically beneficial. Far from it. We are already flooded with AI products causing serious harm, triggering liability claims for negligence and wrongful death, and launching emotional Senate hearings.

In this regard, I have argued that the courts have no factual basis for even defining the purpose of AI training. Although we should not talk about AI as a monolith, the counterpoint to that principle is that it’s generally the same process ingesting the same creative works, whether the AI product is used for scientific research, military applications, medical diagnosis, CSAM, social engineering attacks, or addicting children to establish dangerous “friendships” with machines.

Even if the courts are unwilling to apply such a broad sweep of uncertainty in a copyright context, it is sufficient to say that we have little reason to assume that AI is generally beneficial in the world of creative and cultural production.  And whether the folks at Public Knowledge know it, the courts are at liberty to look beyond the four factors in weighing fair use, especially when they are presented with considerations that have little or no precedent.

It is important to keep in mind that on fair use factor one, the often unwieldy transformative doctrine splits into two distinct branches of case law. The traditional purpose of fair use, dating back to English courts, is to allow new creative expression to flourish, particularly expression that comments upon the work being used. Fair use cases of this nature most often address one user of one work for one clear purpose.

The more contemporary branch of factor one considerations entails mass use of protected works for a technological purpose that can strain against the fair use doctrine. Simply put, fair use was not developed or codified into statute to provide raw materials for technological products, and as discussed in other posts, when the Second Circuit allowed scanning millions of books for Google Books, it stated that the case “tests the boundaries of fair use.” GAI products, whether used for good or ill, lie well outside those boundaries.

Articles like Gallaher’s are not really making a copyright argument but are instead drawing readers to conclude that copyright owners should be required to subsidize AI development whether they like it or not. Other than assuming that Public Knowledge is still a PR firm for Big Tech, I don’t know why an organization with that name takes such a position when countless parents, educators, artists, lawmakers, and medical experts are insisting upon guardrails and oversight for AI in recognition of social harm already being done. This same sober approach must apply to copyright rights and, at the very least, foster a licensing regime that avoids undermining foundational IP principles.

Image source: H9images

Mahmoud v. Taylor:  SCOTUS Marks Insidious Path Toward Book Bans

Mahmoud

In finding for the petitioners in Mahmoud v. Taylor, the Supreme Court’s conservative majority opens another path to banning books in schools—administrative hassle disguised as constitutional principle. The petitioners in the case are three families—one Muslim, two Catholic—with young children in the Maryland Central Public Schools (MCPS) where the board elected to include a number of children’s books with gay or trans characters or subject matter. The families asked the school to accommodate an opt-out for their children, which would entail notifying the families when the books would be used in class and allowing their children to skip those classes without effect on their attendance records.

On the surface, the Court’s finding for the petitioners might seem relatively innocuous. At oral arguments, Justice Alito asked “What’s the big deal?” about allowing families to opt out on religious grounds, and then on June 27, he delivered the majority opinion granting the families a preliminary injunction and thoroughly expressed how the Court would ultimately rule if the case were to proceed.

The big deal about requiring a public school to facilitate an opt-out in this case is that it invites both administrative and pedagogical chaos with the likely result that at least some schools will find it easier to simply keep certain titles out of the classroom. That is, of course, the true goal of whatever group is underwriting the Mahmoud case; and while Alito’s opinion does a reasonable job of camouflaging its own religious bias in constitutional lingo, its errors are hiding in plain sight.

The holding turns substantially on the opt-out question, which is reasonable to a point because compelled conduct by the state can abridge the exercise right in certain circumstances. But here, the opt-out context relies entirely upon the Court’s subjective interpretation of the books at issue, molding the facts to fit the conclusion. More broadly, I believe Mahmoud reflects a generally biased First Amendment jurisprudence that is often too eager to conflate religious “exercise” with religious belief. The two are not the same, either legally or pragmatically.

The Books at Issue

The majority finds that the children’s books in this case “pressure students to conform” to views that conflict with their families’ religious exercise rights. It even describes the books several times as “religiously offensive material,” as if this were a clear and universally applicable fact rather than a subjective opinion. While nobody can doubt that a book presenting homosexuality as “normative” can imply that the religious views of the petitioners are wrong, that consideration is both too broad and too narrow an application of “exercise” at the same time. Too broad because “exercise” cannot encompass every belief in every heart, and too narrow because even other religious exercise demands opposing conduct. For example, in discussing the book Prince & Knight, Justice Alito writes the following:

The book relates that “on the two men’s wedding day, the air filled with cheer and laughter, for the prince and his shining knight would live happily ever after.” Those celebrating the same-sex wedding are not just family members and close friends, but the entire kingdom. For young children, to whom this and the other storybooks are targeted, such celebration is liable to be processed as having moral connotations. If this same-sex marriage makes everyone happy and leads to joyous celebration by all, doesn’t that mean it is in very respect a good thing?

On that basis, consider the Episcopalians who, in my town, light up their church every June for Pride while the churches of other denominations do not. Suppose an Episcopalian family in our public school sincerely believes, under this Court’s reasoning, that a children’s book depicting the joyous celebration of a man and woman getting married promotes the view that same sex marriages are morally wrong. That interpretation may appear irrational, but it is identical in logic to Alito’s description above—unless, of course, we allow that the narrow, religious bias inherent to his interpretation is constitutionally sound. Of course it is not.[1]

In a concurring opinion, Justice Thomas highlights the Court’s religious bias by stating that the school “…rather than attempt to ‘weave the storybooks seamlessly into ELA lessons,’ the Board could cabin its sexual- and gender-identity instruction to specific units.” But that reasoning only makes sense to those who insist that gay and trans characters, like the real people they represent, must remain sequestered from everyday American life in order to avoid offending people.[2]

Thus, the Court is blind to the fact that it recommends accommodation for any family claiming religious exercise, even to opt out of what I assume the majority would struggle to describe as “offensive to religious exercise.” And because it would strain logic to square its reasoning, the Court exposes its own religious preferences while feigning a neutrality the Constitution requires. This blinkered view is emphasized by finding that the 1972 case Yoder v. Wisconsin is almost binding precedent for the result in Mahmoud.

Yoder v. Wisconsin is Inapt

In Yoder, the petitioners, who were Amish, sought an exemption to Wisconsin’s law requiring that all residents attend school through the age of 16. The plaintiffs, concerned about many aspects of public high school that conflicted with their religion, won the right to allow their adolescent children to opt out of the school system on the basis that the state law imposed a heavy burden on their religious exercise. I agree with the result in Yoder, but not without acknowledging the compelling fact that the Amish are a unique society within American society, which makes them highly distinguishable from the parents in Mahmoud.

Most importantly, the Amish did not want it both ways. They did not seek to alter a single aspect of the public-school administration or curriculum; they simply wanted their children excused from compulsory attendance. By contrast, the parents in Mahmoud—and all parents similarly situated—want to remain in the public school while demanding a degree of impractical accommodation for their individualized religious beliefs. That the majority overlooks this chasm of distinction between the two cases is consistent with its willful blindness to the pedagogical and administrative burdens its holding fosters.

Notably, the Court fails to recognize that, as a legal matter, the plaintiffs in Yoder sought the equivalent of moving children from public to religious school. This elision of reasoning then allows the Court to opine that the option of religious schools for the Mahmoud parents would be too costly—a consideration that does not sound in Yoder or the Constitution, and one the Court fails to balance with the burden on the school to accommodate any family with any stated belief offended by the curriculum.

Banning Books is Easier than Administrative Hassle

A classroom environment that is welcoming to all students is something to be commended, but such an environment cannot be achieved through hostility toward the religious beliefs of students and their parents.

Nothing in the record of this case justifies the word “hostility” in that sentence, and yet it is a telling choice—one that demonstrates this Court’s willingness to step outside its purview and advocate on behalf of some (though certainly not all) sincerely held beliefs. People who want to be recognized for who they are—gay, trans, etc.—are not demonstrating “hostility” toward religion by that act alone. And while we must admit that hostility is inevitable when views, beliefs, and religions collide, these social animosities are not reconciled by the Court finding without reason that one American’s mere existence is “hostile” to another American’s religion.

Meanwhile, within the scrum and squabble of American life, the public school is obliged to include materials that present the world as it is, not the world that certain parties wish it to be. Yet, the Court states, “[The books] are clearly designed to present certain values and beliefs as things to be celebrated  and certain contrary values and beliefs as things to be rejected.” All media conveys a point of view, and all media are subject to viewer interpretation. The first grader’s introduction to cosmology will categorically reject the views of the creationist, and this reference to science is well-founded because a book depicting gay or trans people as “normative” is a matter of scientific reality.

But under the Court’s reasoning in Mahmoud, shall we divide the class to learn about Galileo the scientist in one room and Galileo the heretic in another? Or shall the public school not merely allow a student to be shielded from proven science or history, but also advance him through the grades when he produces incorrect answers in light of his sincerely held beliefs? In Justice Sotomayor’s dissent, she summarizes the problem thus:

Given the great diversity of religious beliefs in this country, countless interactions that occur every day in public schools might expose children to messages that conflict with a parent’s religious beliefs. If that is sufficient to trigger strict scrutiny, then little is not.

Exactly. The broad spectrum of books, ideas, and facts that may be presented in school, even in the K-5 years, will inevitably contain some material anathema to some family’s sincerely held religious beliefs. In that light, public schools cannot reasonably be burdened with managing a dynamic rubric, tracking which families may opt out from specific lessons and on what basis. As Justice Sotomayor states, “Many school districts, and particularly the most resource strapped, cannot afford to engage in costly litigation over opt-out rights or to divert resources to tracking and managing student absences.”

Thus, in recognizing the impracticality of a complex opt-out policy, the alternative choices would be to allow ad hoc opt-outs without explanation or to remove certain materials from the curriculum, which is undoubtedly the goal of the lawsuit in Mahmoud. Of course, this Court would never endorse all sincerely held beliefs under its reasoning.

An Extreme Hypothetical to Make the Point

I have never liked the Pledge of Allegiance. I consider it a creepy, un-American act of performative patriotism, and the words “under God” not only conflict with my sincerely held beliefs but also undermine that next word “indivisible.” Like many students, I recited the Pledge as a young child, mumbled it as I got older, and didn’t say it at all by the time I was a teenager. But as a parent, knowing every public school would maintain the ritual, should I have sought an opt-out for my children, demanding on the basis of my First Amendment rights that my kids should have been allowed to be tardy every day to avoid mere exposure to the Pledge?

I doubt any court would support that claim, even under the ruling in Barnette (1943), also cited in Mahmoud. There, the Court found for Jehovah’s Witness petitioners who objected to a West Virginia state rule compelling students to salute the flag during the Pledge of Allegiance. The Court agreed with the petitioners’ religious right not to worship a graven image, though of course, the compelled salute also offends the speech right, a broad view of the exercise right, potentially the redress right, and is just plain offensive. But just as Yoder is inapt in Mahmoud, so too is Barnett inapt in my hypothetical desire to have my kid opt out of every morning to avoid the Pledge.

Unless the school compels a specific action other than simply being in the presence of the “offensive material,” the impracticality of my request to allow my kids to be tardy every day should outweigh my personal belief that exposure to the Pledge “harms” my First Amendment right to religious—in this case irreligious—instruction of my children. As stated, sowing impracticality is arguably the aim and result of the Mahmoud case—to implicate so much administrative difficulty for at least some schools that books containing the subject matter at issue are simply removed from the shelves.

A Tradition of Bias in “Exercise” Jurisprudence?

“The dissent sees the Free Exercise Clause’s guarantee as nothing more than protection against compulsion or coercion to renounce or abandon one’s reli­gion,” states the Court. I believe the dissent is right—or should be.

To be clear, I would demote my own “religious” beliefs if First Amendment jurisprudence remained narrowly tailored to “exercise” under a strict textual interpretation. I freely admit that as an atheist, I do not engage in what any ordinary person would call religious “exercise” in the sense that my friends attend places of worship and observe certain rites and practices. In this regard, my sense is that conservative jurisprudence tends to want to encompass belief (though not every belief), which is subjective and intangible in contrast to “exercise,” which entails demonstrable conduct.

While it is reasonable that where the state compels certain conduct, the courts must consider whether such compulsion is an abridgement of “exercise.” But with the possible exception of the Amish and truly cloistered communities, this principle cannot apply to mere exposure to ideas, views, expressions, or events that are inescapable realities of living in a polyglot democracy. Public schools sit squarely in the center of public life, and in school as in the broader community, tolerance of even the offensive is the foundation of domestic tranquility. The family that feels otherwise is not only free, but I would argue obligated, to choose an educational alternative that comports with their chosen forms of intolerance.

Conclusion

The Court’s holding in Mahmoud v. Taylor is not surprising, though I admit I was hopeful that Justice Barrett, who has revealed herself an independent thinker, might have written a nuanced concurrence. Instead, the majority’s opinion offers much to justify those who view the current Court as warped by theocratic sentiment that comes dangerously close to advancing a view of “exercise” that would swallow the establishment clause. It speaks in the language of religious neutrality but articulates a clear preference for certain religious beliefs over others.

[1] Further, Alito’s reference to the joy of the “entire kingdom” is simply bizarre. Does he mean to suggest that if some subjects were illustrated as unhappy, perhaps wearing crosses and unhappy, that the book would no longer “pressure conformity” as the Court maintains?

[2] It is curious how often Justice Thomas expresses a reasoning that many Americans would apply to reject the validity of his own interracial marriage.

America’s AI Action Plan:  Strength Through FAFO

ai

With an introduction that combines the quasi-erudition of Big Tech utopianism and just enough dipshit to sound like Donald Trump, the White House unveiled an action plan on artificial intelligence that is part magical thinking and part policy statement. It all adds up to one bottom line: “Let Big Tech do what it wants, and things will be great.”

In fairness, letting Big Tech do what it wants has a solid bipartisan tradition in U.S. policy, but the naivete of the late 1990s is ancient history. The public and Members of Congress are now well versed in the many negative consequences of laissez faire policy that allowed Web 2.0 to run amok liability free, and yet, despite hearing after hearing with Senators proclaiming outrage at the tech giants, we are poised to approach the development of artificial intelligence with the same doe eyed, babes-in-the-woods innocence of people who cannot learn from experience.

Regarding most of the action plan’s content, the language itself describing each initiative—from science and medicine to national defense—is innocuous and irrelevant. Because no matter what is proclaimed as a goal, the question of AI development that can benefit the American people comes down to guardrails and oversight—and usually some combination of the two. Unfortunately, we have neither.

The tech companies have repeatedly demonstrated that they possess no ethics that would prioritize public interest over profit; the Trump administration has no credibility about anything; and Congress has largely been reduced to performance art, promising for the last eight years that it will finally “rein in tech.” Thus, the AI action plan, like so many plans of the current administration, boils down to fuck around and find out (FAFO).

Unlike major transformative undertakings of the past—the action plan cites the space race—no technology has had the potential to crawl into every aspect of social, economic, and political life as AI. Data runs the world, which means those who control the data run the world. Hence, the promise of what AI could do for society rests entirely on the guardrails and oversight which the industry rejects out of hand and the Trump administration believes are unnecessary.

As one obvious example, a whole section of the plan discusses “empowering American workers,” which is boilerplate for any administration except that most predictions about AI include job loss at unprecedented scale. To navigate this fresh terrain—either to mitigate job loss or address the consequences of job loss at scale—requires leadership that is intelligent and gives a damn about people. But this administration is neither intelligent nor cares about anyone, including the people who voted for them.

Consider one example from a friend who is sanguine about AI’s potential, and who recently mentioned the likelihood that an AI will soon be much better at reading a medical scan or X-ray than a human radiologist. That prediction, and the penumbra of medical advancements it implies, is entirely reasonable but also invites a cascade of ethical considerations that the party of Trump would sweep into the dustbin called “over regulation.”

So, sure, in theory we could be the beneficiaries of faster, more accurate, and cheaper diagnostics with the potential to alleviate scenes like the bottleneck seen every day in my local ER. But ordinary Americans will not truly benefit from these and other promised advancements without public oversight, and I imagine the party that wants to kick millions of Americans off health insurance rolls and scale back essential services doesn’t give a flying fuck.

As the mantra about Web 2.0 preaches, “If the service is free, then you’re the product.” That was and remains a dangerous consequence of social media platforms. But what happens when the same principle applies to access to medical or other critical services wholly controlled by one or two AI companies—say Meta or Amazon?—run by leaders with zero morals? Maybe the out-of-pocket price comes down initially and revitalizes the utopian “age of abundance” rhetoric from tech’s cheerleaders, but the real implication of having a handful of companies providing essential services is technological feudalism. And that prospect undermines the animating imperative of the AI action plan—i.e. to “beat China” in this new cold war.

But at present, we’re just too stupid to beat China at this game because we’ve reached a crisis state when we have no fucking idea who we are as a nation anymore. One section of dipshittery in the action plan states that AI will “protect speech and American values” and that these ends will be achieved, in part, by “eliminat[ing] references to misinformation, Diversity, Equity, Inclusion, and climate change.” Right. Fuck the best and brightest for the sake of the vested whitest. Because that’s how we won the space race, right? Without mathematicians Katherine Goble Johnson, Dorothy Vaughn, and Mary Jackson.

Meanwhile, the White House built on misinformation, sedition, hate, crime, grift, ignorance, and authoritarian tactics has a plan to “protect American values” by behaving exactly like China and other anti-democratic societies. I’d call that burning down the village to save it, but especially with AI, it’s actually erecting a Potemkin Village and calling it America. Web 2.0 has been a multipronged disaster because congressional leaders on both sides of the aisle naively decided to let the experiment run for years until finally expressing regret circa 2017. Now, we are poised to double down on the errors of the 1990s and FAFO with a tech fraught with uncertainties for the American people. Beat China? We’re not smart enough to meet the moment.

Photo by Yacobchuck