The AI revolution has not crept in quietly -- it has detonated!

We have arrived at a point where Artificial Intelligence can generate music, write software, produce photographs and artwork, create video, draft legal briefs, and increasingly produce entire bodies of work from nothing more than human direction. The debate surrounding AI and its ability to function as a "creator" of Intellectual Property (IP) is no longer theoretical -- it is playing out in Copyright Offices, courtrooms, and ethics committee chambers simultaneously, and the law is struggling to keep pace.

I raise this because I have been listening to songs created and "performed" entirely by AI, and the experience made me stop. Two examples worth your time:

Michael Bennett Moves Everyone to Tears with "Echoes of a Father's Heart" | America's Talent Show
https://youtu.be/oHZ7NZ9hP3g?is=oQ_jLpDn0xTkVVKF

NORTH WIND CALLS
https://youtu.be/blnRn30okOk?is=j0xyvBB8krNlbDNN

Both were created and "performed" by AI. Yet behind each digital performer stands a human being -- someone who wrote the lyrics, shaped the direction, composed the prompts, selected among outputs, and guided the model toward a finished result. That human-machine collaboration sits at the center of a legal debate that has been addressed, but only partially, and the gaps left behind are widening by the day.

The Copyright Conundrum

To understand where the law stands, consider a progression.

An artist paints a portrait by hand. She chooses the subject, the composition, the palette, the placement of every shadow and highlight. Human authorship is unambiguous.

Now consider a digital artist using Photoshop -- digital brushes, filters, generative fill, AI-assisted masking. The human still makes substantial creative decisions, but the machine is executing some of the expressive work.

Now consider someone who types: "Create a photorealistic portrait of an astronaut standing on Mars at sunset" -- and an AI generates the entire image, making thousands of expressive decisions about lighting, facial structure, composition, texture, and atmosphere that the human never specified.

Who created that image? The human who selected the concept and the words? The machine that determined every visual element? Both? Or -- under the current legal framework -- neither?

This is not a hypothetical. The U.S. Copyright Office has been working through exactly this question in cases like Thaler v. Perlmutter (No. 1:22-cv-01564, D.D.C. 2022) and the now-landmark Zarya of the Dawn registration dispute. In Zarya of the Dawn, artist Kristina Kashtanova created a comic book using Midjourney. She did not simply prompt the AI and walk away -- her process involved human-written text, iterative prompt engineering over hundreds of intermediate images, and deliberate selection, cropping, and arrangement of the resulting visual elements. The Copyright Office initially registered the work, reconsidered after learning the full scope of Midjourney's involvement, and ultimately issued a limited registration covering only the portions attributable to human authorship. The Midjourney-generated images themselves -- the visual expression -- were excluded.

That distinction is the foundation of the current framework. The Copyright Office's position, reaffirmed in its 2023 registration guidance, draws four operative lines:

No machine authors. Courts and the Copyright Office maintain that the Copyright Act protects only human-created works.
Prompts do not count. Simple text prompts function as unprotectable ideas rather than creative control.
Mixed contributions. Only the human-authored portions of an AI-assisted project are eligible for copyright registration.
Public domain. Fully autonomous AI outputs lack an eligible author and remain unprotected.

The Copyright Office is not saying that artists cannot use AI. It is saying something considerably more precise: using AI as a tool does not eliminate copyright protection, but the human being must actually contribute sufficient creative expression to the resulting work. Selecting an output is not the same as authoring it.

That distinction was defensible when AI was a blunt instrument requiring substantial human refinement. It is becoming increasingly difficult to maintain as AI systems grow capable of generating the actual expressive substance of a work from a handful of words. We are moving -- rapidly -- from AI-assisted creation, where the human does the creative work and the machine executes it, toward AI-mediated creation, where the machine determines the expressive details and the human curates the result. The law has not yet drawn a line between those two states, and the technology is not waiting for it to do so.

Zarya of the Dawn was an early warning. AI-generated music is another. Generative photography, illustration, video, and multimodal systems will make the questions harder still.

The Courtroom Problem

The copyright question, urgent as it is, may not even be the most immediate problem. A parallel crisis is unfolding not in the Copyright Office but in the courtroom itself -- and it concerns not who owns AI output, but who gets to use AI as a tool in litigation.

In March 2026, U.S. Magistrate Judge Maritza Dominguez Braswell issued a ruling in Morgan v. V2X, Inc. (Case No. 1:25-cv-01991, D. Colo.). The court ordered both parties in a federal employment lawsuit to stop uploading confidential information into consumer AI platforms. The defendant -- V2X, a defense contractor -- sought a protective order barring AI tools from receiving confidential information unless the provider was contractually prohibited from using inputs for training and required to allow their deletion. The self-represented plaintiff objected, arguing the restrictions would create an unfair "technological gap" by barring him from modern analytical tools while the defendant's lawyers retained access to proprietary, enterprise-grade AI systems with precisely the contractual protections the order demanded.

The court acknowledged the disadvantage. Then granted the order anyway.

The protective order was facially neutral -- it applied to both sides. But it was not economically neutral. V2X has a legal department, outside counsel at firms with procurement teams, information-security officers, and the institutional budget for enterprise AI contracts with robust confidentiality terms. The plaintiff had a consumer subscription and no lawyer. The court banned the one tool the plaintiff could realistically access, permitted the tier he could not afford, noted the problem in a footnote, and moved on.

That is not a technological gap. It is a due process gap wearing technological clothing.

The Fairness Doctrine in Modern Dress

This is the Fairness Doctrine problem -- not as a historical reference, but as a structural diagnosis.

The original Fairness Doctrine (FCC, 1949) held that broadcasters controlling a scarce public resource -- the broadcast spectrum -- bore affirmative obligations to serve the public interest rather than institutional advantage. It was repealed in 1987 on the premise that a competitive marketplace would self-correct. The AI legal market is demonstrating, in real time, that it will not.

The scarce resource here is not spectrum. It is procedural parity: the ability to use equivalent analytical tools in litigation. When a corporation like V2X has enterprise AI fully integrated into its legal workflow -- AI-assisted brief drafting, document review, legal research, and strategic analysis -- and a self-represented plaintiff cannot use consumer AI without risking court sanctions, the disparity is not incidental. It is structural. The individual plaintiff has to choose between violating a protective order and doing all of that manually against a fully resourced opponent.

Large firms can deploy procurement teams, information-security officers, and vendor counsel to negotiate data-processing terms, audit rights, and retention windows that solo practitioners and self-represented litigants simply cannot reach (Leben, 2026). The market has not produced a floor -- it has produced a ladder with the bottom rungs removed.

Enterprise AI tiers are priced for institutional procurement. Consumer tiers carry the confidentiality risk profiles that ethics guidance increasingly treats as disqualifying. The middle is thin -- there is no affordable tier that simultaneously satisfies the highest bar of ethics guidance and the financial constraints of someone representing themselves in a Federal employment dispute. A rule that is formally neutral produces systematically unequal outcomes based on who can afford access to the infrastructure it assumes.

This also implicates a deeper problem in legal ethics that scholars like Steve Leben (Harvard JOLT, 2026) have begun to articulate: the profession has been reading Rule 1.6's confidentiality requirements as if enterprise-grade security is the minimum standard every lawyer must meet, when the rule's actual text -- and its Comment 18 factors, which explicitly include cost, difficulty, and effect on representation -- supports a more calibrated analysis. A "reasonable efforts" standard should not quietly become an "enterprise required" standard. But courts issuing orders premised on that miscalibration are entrenching the inequality, and those orders fall hardest on the people who were already losing.

The Absence of a Regulatory Fix

The original Fairness Doctrine had an enforcement mechanism: the FCC could revoke broadcast licenses. There is no equivalent here.

Bar associations issue ethics opinions with no enforcement power in individual cases. Courts issue protective orders case by case. Congress has not acted on AI and legal access -- while legal AI products like Westlaw's CoCounsel and Counsel AI Corporation's Harvey AI continue to scale exclusively within institutional markets, priced and architected for the clients who need them least. The institution best positioned to address this disparity -- the Judiciary -- is instead producing decisions like Morgan that entrench it, while acknowledging in footnotes that they are doing so.

This is the regulatory void: a problem clearly visible, openly admitted, and structurally unaddressed.

The Question the System Is Not Asking

The copyright question and the access-to-justice question are, at their core, the same question expressed in different legal vocabularies: who benefits from AI, and who bears its costs?

In the copyright context, the question is whether meaningful human authorship exists -- and who gets to claim the creative product of AI-assisted or AI-mediated work. In the litigation context, the question is whether meaningful procedural parity exists -- and who gets to use the analytical power of AI-assisted or AI-mediated legal work.

In both contexts, the answer currently skews toward the institutionally advantaged.

As large firms pour resources into enterprise-grade AI, how will a self-represented litigant keep up? This is a question the justice system has yet to answer -- and what is clear is that the system is not asking it with any urgency, because the people most harmed by the non-answer are the same people who have always had the least voice in shaping legal doctrine: individuals without lawyers, without money, and without the institutional standing to make their procedural disadvantage legible as a civil rights issue.

Technological inequality is no longer limited to who has a faster computer or a better smartphone. It is beginning to determine who can effectively participate in the legal, economic, and creative systems being built around Artificial Intelligence. If we are not deliberate in how we design the rules that govern AI access, the revolution will not simply create a new class of creators and a new class of litigants. It will create a new class of people who can afford to have machines work for them -- and a much larger class who cannot.

The Fairness Doctrine was repealed because regulators decided the market would handle it.

The legal AI market is demonstrating, in real time, that it will not.

REFERENCES

Analla, T. (2023, March 6). Zarya of the Dawn: How AI is Changing the Landscape of Copyright Protection. Harvard Journal of Law & Technology. https://jolt.law.harvard.edu/digest/zarya-of-the-dawn-how-ai-is-changing-the-landscape-of-copyright-protection

Chesney, B., & Citron, D. K. (2019). Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security. Berkeley California Law Review, Volume 107(Issue 6), 1753–1820. https://lawcat.berkeley.edu/record/1136469?v=pdf | https://doi.org/10.15779/Z38RV0D15J

Federal Register. (2023, March 16). Federal Register/Vol. 88, No. 51/Thursday, March 16, 2023 / Rules and Regulations. The Federal Register. https://www.govinfo.gov/content/pkg/FR-2023-03-16/pdf/2023-05321.pdf

Knowing Machines. (2025, November 21). Registration of "Zarya of the Dawn" Registration # VAu001480196 (U.S. Copyright Office) Registration canceled: 2/21/2023. Knowing Machines. https://knowingmachines.org/knowing-legal-machines/legal-explainer/cases/registration-of-zarya-of-the-dawn

Lawler, R. (2023, February 23). The US Copyright Office says you can't copyright Midjourney AI-generated images. The Verge. https://www.theverge.com/2023/2/22/23611278/midjourney-ai-copyright-office-kristina-kashtanova

Leben, S. (2026, July 10). AI, Confidentiality, and the Stratified Legal Profession. Harvard Journal of Law & Technology. https://jolt.law.harvard.edu/digest/ai-confidentiality-and-the-stratified-legal-profession

Morgan v. V2X, Inc., 1:25-cv-01991. (2025, June 26). CourtListener. https://www.courtlistener.com/docket/70646477/morgan-v-v2x-inc/

Skladany, M. (2026, June 6). daptive AI Laws: To Address the AI Revolution, Law Needs One of Its Own. Harvard Journal of Law & Technology. https://jolt.law.harvard.edu/digest/adaptive-ai-laws-to-address-the-ai-revolution-law-needs-one-of-its-own

Thaler v. Perlmutter (No. 1:22-cv-01564). (2022, June 2). Courtlistener: Case No. : 1:22-cv-01564 - IN THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF WASHINGTON D.C. Stephen Thaler, an individual Plaintiff, v. Shira Perlmutter, in her official capacity as Register of Copyrights and Director of the United States Copyright Office; and The United States Copyright Office; Defendants. Courtlistener. https://storage.courtlistener.com/recap/gov.uscourts.dcd.243956/gov.uscourts.dcd.243956.1.0.pdf

U.S. Congress. (2025, July 18). Generative Artificial Intelligence and Copyright Law | Congress.gov | Library of Congress. Congress.gov | Library of Congress. https://www.congress.gov/crs-product/LSB10922

U.S. Copyright Office. (2023, February 21). 2023.02.21 Zarya of the Dawn (Registration # VAu001480196) [Letter]. https://www.copyright.gov/docs/zarya-of-the-dawn.pdf

U.S. Copyright Office. (2023b, March 16). Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence. https://www.copyright.gov/ai/ai_policy_guidance.pdf

U.S. Copyright Office. (2026). Copyright and Artificial Intelligence. Copyright and Artificial Intelligence | U.S. Copyright Office. https://www.copyright.gov/ai/

Wolfson, S. (2023, November 8). Zarya of the Dawn: US Copyright Office Affirms Limits on Copyright of AI Outputs - Creative Commons. Creative Commons. https://creativecommons.org/2023/02/27/zarya-of-the-dawn-us-copyright-office-affirms-limits-on-copyright-of-ai-outputs/