Introduction
Few doctrines in American law generate more confusion, more litigation, and more commercially consequential uncertainty than the fair use doctrine. Codified at 17 U.S.C. § 107 of the Copyright Act, fair use is a defense to copyright infringement that permits the use of a copyrighted work without the copyright owner’s permission under certain circumstances. It is not a license, not a right, and not a guarantee. It is a statutory affirmative defense — meaning that the burden falls on the party claiming fair use to prove that their use qualifies, not on the copyright owner to prove it does not.
For US businesses, the fair use doctrine is simultaneously a shield and a source of risk. As a shield, it permits companies to engage in commentary, criticism, parody, research, education, and transformative creative activity without obtaining permission for every use of copyrighted material. As a source of risk, it creates a zone of legal uncertainty where well-intentioned businesses can find themselves defendants in costly infringement litigation, and where the outcome of any given case can be genuinely difficult to predict. Understanding the doctrine — not just its four statutory factors but the nuanced, context-dependent way courts actually apply them — is essential for any business that creates content, uses third-party materials, builds technology platforms, or engages in commentary about competitors, public figures, or cultural phenomena.
This article provides a thorough examination of the fair use doctrine as it operates in US law. It explains the historical origins of the doctrine, analyzes each of the four statutory factors with reference to key case law, addresses the specific commercial contexts in which fair use issues most commonly arise for businesses, and offers practical guidance on how to assess fair use risk and structure activities to minimize exposure to infringement claims.
Origins and Constitutional Basis
Fair use is not a modern invention. Its roots lie in the equitable doctrine developed by English courts in the eighteenth century and subsequently adopted by US courts well before the doctrine was codified by Congress. The Supreme Court recognized fair use as a common law doctrine as early as Folsom v. Marsh (1841), in which Justice Story articulated a multi-factor analysis — considering the nature and objects of the selections made, the quantity taken, and the degree of prejudice to the original work — that prefigured the modern statutory framework. Congress formally codified the doctrine in the Copyright Act of 1976, enacting what is now 17 U.S.C. § 107.
The constitutional grounding of fair use is significant and often overlooked. The First Amendment to the Constitution guarantees freedom of speech and freedom of the press. Copyright law, which grants private parties monopoly control over expressive works, creates an inherent tension with these First Amendment values. Fair use is one of two primary doctrines through which US copyright law accommodates this tension — the other being the idea/expression dichotomy, which ensures that copyright protects only the particular expression of an idea, not the idea itself. The Supreme Court has described fair use as a “safety valve” built into copyright law that prevents it from functioning as a tool for censoring commentary, criticism, and public discourse about copyrighted works. This constitutional dimension is relevant to how courts approach fair use claims: they tend to be more receptive to fair use arguments that serve speech and public interest values and more skeptical of fair use claims that serve purely commercial ends with little broader social benefit.
The Four Statutory Factors
Section 107 of the Copyright Act sets out four non-exclusive factors that courts must consider in evaluating a fair use claim. The statute expressly states that these factors are to be weighed “in light of” the preamble examples it provides — criticism, comment, news reporting, teaching, scholarship, and research — although courts have long recognized that fair use is not limited to these categories. The factors are not a checklist; they are examined holistically, and no single factor is automatically determinative. A finding that any one factor favors or disfavors fair use does not resolve the analysis. The Supreme Court reaffirmed this holistic approach most recently in Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith (2023), while also importantly reemphasizing the weight courts should give to the commercial nature of the use.
Factor One: The Purpose and Character of the Use
The first factor asks what the defendant did with the copyrighted work and why. Courts examine two principal dimensions: whether the use is commercial or noncommercial in nature, and whether the use is “transformative.” The Supreme Court’s landmark 1994 decision in Campbell v. Acuff-Rose Music, Inc. established that the concept of transformation — whether the new work adds something new, with a different purpose or character, and does not merely supersede the objects of the original — is the central inquiry within this factor. A use is transformative if it alters the original with new expression, new meaning, or new message, rather than simply reproducing it for the same purpose the original served.
Transformativeness has, in the years since Campbell, become the most heavily weighted consideration in fair use analysis, to the point where some commentators argued it had come to dominate the entire four-factor test. That trend was checked, at least in part, by the Supreme Court’s 2023 decision in Andy Warhol Foundation v. Goldsmith. In that case, the Court held that Andy Warhol’s silkscreen images derived from photographer Lynn Goldsmith’s photograph of the musician Prince were not fair use when the Warhol Foundation licensed them commercially to magazines for the same basic purpose — illustrating articles about Prince — as the original photograph. The Court’s majority opinion, written by Justice Sotomayor, emphasized that the first factor requires courts to look not merely at whether a work is “transformative” in an aesthetic sense, but at whether the secondary use serves a “sufficiently distinct” purpose or function from the original. Where a secondary work competes with the original in the same market for the same purpose, the transformative nature of the artistic style or medium is insufficient to support a fair use finding.
The commercial nature of a use does not automatically defeat fair use — as the Court noted in Campbell, nearly all secondary uses are commercial in some sense. But commercial uses are more likely to weigh against fair use, particularly when the secondary work directly substitutes for the original in a commercially competitive market. Noncommercial uses, including educational, research, and personal uses, are viewed more favorably under this factor, though they do not receive a pass on the remaining factors.
Other purposes that courts have found to favor fair use under the first factor include: criticism and commentary on the original work (quoting from a book in order to critique it), parody (using an original work to mock or comment upon that very work, as distinguished from satire which uses a work to comment on unrelated subjects), news reporting (reproducing a photograph or video clip in the context of news coverage), and certain transformative artistic uses (creating a new work that fundamentally reimagines the original for a different purpose and audience).
Factor Two: The Nature of the Copyrighted Work
The second factor examines the characteristics of the original work that was copied, not the secondary work. Two dimensions are relevant. First, courts distinguish between creative or expressive works and factual or informational works. Copyright protection is stronger for highly creative works — poetry, fiction, music, fine art — than for works whose value lies primarily in the factual information they contain, such as databases, directories, or news articles. Reproducing portions of a factual work therefore tends to be viewed more favorably under this factor than reproducing portions of a purely creative one.
Second, courts consider whether the original work has been published or remains unpublished. The Copyright Act provides heightened protection for unpublished works, reflecting the author’s interest in controlling the timing and manner of first disclosure. The Supreme Court’s decision in Harper & Row Publishers, Inc. v. Nation Enterprises (1985) held that The Nation magazine’s publication of excerpts from President Gerald Ford’s unpublished memoirs — obtained without authorization — was not fair use, with the unpublished nature of the work weighing heavily against the defense. This principle remains important: the fair use defense is considerably more difficult to sustain when the work copied has not yet been made public.
In practice, the second factor rarely drives the outcome of a fair use case. It tends to be treated as a tiebreaker rather than a primary determinant, and it most clearly matters at the extremes: where the work is highly creative and unpublished, this factor weighs strongly against fair use; where the work is predominantly factual and published, it weighs in favor.
Factor Three: The Amount and Substantiality of the Portion Used
The third factor considers both the quantitative and qualitative dimensions of what was taken. Quantitatively, courts look at what proportion of the original work was reproduced. Taking a small percentage of a large work is generally more defensible than taking a large percentage of a short one, though no safe harbor based purely on word count or percentage exists. The Copyright Office and various courts have at times been asked to establish a “safe” number of words or seconds of music that can be taken without permission — no such numerical threshold has ever been accepted as a legal standard. Any claim that “up to 400 words” or “up to 30 seconds” is automatically permissible is legally unfounded.
Qualitatively, courts ask whether the defendant took the “heart” of the original work — the most memorable, distinctive, or commercially valuable portion. In Harper & Row, the Court found that The Nation had reproduced only about 300 words of a memoir running to more than 200,000 words, but that those 300 words were the most newsworthy and compelling portions of the book. The qualitative significance of what was taken weighed against fair use even though the quantity was modest in percentage terms. Conversely, in cases where a defendant needed to reproduce more of a work in order to accomplish a clearly legitimate purpose — such as a search engine creating thumbnail images for an index — courts have found that the quantity taken was no greater than necessary, which mitigates the weight of this factor.
The key principle underlying the third factor is proportionality: a defendant who takes only what is necessary for the legitimate purpose articulated under the first factor is in a stronger position than one who reproduces more than necessary. This principle has practical implications for businesses: using the minimum amount of copyrighted material necessary to accomplish the stated purpose, rather than reproducing entire works or large portions when less would suffice, strengthens a fair use argument and demonstrates good faith.
Factor Four: The Effect on the Potential Market for the Original Work
The fourth factor — the effect of the use upon the potential market for or value of the copyrighted work — has historically been described by courts as the most important of the four factors, a view expressed by the Supreme Court in Harper & Row and reaffirmed in subsequent decisions. The inquiry is not limited to actual harm already sustained by the copyright owner; it extends to the potential market, including markets the copyright owner has not yet entered but into which the copyright owner might reasonably be expected to license the work.
Courts ask whether the secondary use harms both the market for the original and the market for derivative works. If widespread conduct of the type engaged in by the defendant would materially damage the market for the original — either by directly substituting for it or by reducing the copyright owner’s ability to license the work — this factor weighs heavily against fair use. If, on the other hand, the secondary use serves a fundamentally different market from the original and does not act as a substitute — a critical review does not substitute for the reviewed book; a parody does not substitute for the original song — this factor is more likely to favor the defendant.
The fourth factor also distinguishes between harm caused by market substitution and harm caused by criticism or commentary. A negative book review that reduces sales of the reviewed book causes market harm, but it is harm caused by the critical speech, not by market substitution. Courts do not treat this type of harm as relevant under the fourth factor; only substitutional harm — where the secondary use takes the place of the original or its licensed derivatives in the market — counts against the defendant. This distinction is important for businesses that engage in comparative advertising, competitive commentary, or critical reviews of competitors’ products: the fact that such activity may damage a competitor’s sales does not mean it fails the fourth factor test.
Fair Use in Specific Commercial Contexts
Commentary, Criticism, and Parody
Commentary and criticism of copyrighted works are among the most clearly protected categories of fair use. A book reviewer, film critic, or technology journalist who quotes from the work under review in order to illustrate or support their critical assessment is engaging in a use that the fair use doctrine was specifically designed to protect. The same principles apply to businesses that engage in honest comparative commentary about competitors’ products or services, or that critically analyze publicly available information about others in their industry.
Parody — defined as a work that comments on or criticizes the original by imitating it in an exaggerated or absurd fashion — receives special treatment under fair use doctrine following the Supreme Court’s decision in Campbell v. Acuff-Rose. The Court held that parody inherently requires some degree of reproduction of the original work, because the audience must be able to recognize what is being parodied. A parodist who reproduces just enough of the original to “conjure up” the original for the audience, while using that reproduction for the new, critical purpose, is in a strong fair use position. Crucially, the Court distinguished parody from satire: a work that uses a copyrighted song merely as a vehicle for commentary on unrelated social or political issues (satire) does not require appropriation of that particular copyrighted work and therefore receives less solicitude under fair use doctrine. For businesses considering using third-party creative works in advertising or brand communications for satirical effect, this distinction is commercially significant.
News Reporting and Documentary Content
The fair use doctrine has historically accommodated the use of copyrighted material in the context of news reporting and documentary filmmaking, recognizing the public interest in accurate and thorough journalistic coverage of events. A news broadcast that shows a few seconds of a copyrighted video in the course of reporting on the event depicted, a documentary filmmaker who includes a copyrighted song playing in the background of footage shot at a public event, or a journalist who reproduces a quoted passage from a published document to illustrate a news story — all of these uses are likely to be evaluated favorably, particularly under the first and fourth factors. However, news and documentary contexts do not confer blanket immunity. Courts have denied fair use in news contexts where the reproduction was excessive relative to the reporting need, where the material was used primarily for its entertainment value rather than its informational content, or where the use clearly displaced potential licensing income.
Technology: Search Engines, AI, and Data Processing
Some of the most consequential fair use disputes of the past two decades have arisen in the technology sector, where the application of old copyright principles to new technologies has produced significant legal uncertainty. Two decisions deserve particular attention.
In Kelly v. Arriba Soft Corp. (9th Cir. 2003) and the related Perfect 10, Inc. v. Amazon.com, Inc. (9th Cir. 2007), the Ninth Circuit held that a search engine’s creation of thumbnail images of copyrighted photographs for use in search results constituted fair use. The court found the use transformative because the thumbnails served a different function from the originals — an index and reference tool rather than a high-quality image for aesthetic enjoyment — and because the thumbnails did not substitute for the originals in their primary market. These decisions established important precedent for technology platforms whose core function involves processing and indexing copyrighted material at scale.
The Supreme Court’s decision in Google LLC v. Oracle America, Inc. (2021) extended fair use protection to Google’s copying of approximately 11,500 lines of Oracle’s Java API — the declaring code and structure necessary for programmers familiar with Java to write programs compatible with Android. The Court, applying all four factors, found Google’s use fair: the copying was transformative in that it was used to build a new smartphone platform serving a different market from Oracle’s original Java platform; the amount taken was limited to what was necessary for interoperability; and the relevant market evidence did not clearly show that Google’s use harmed Oracle’s potential licensing market in the smartphone context. Google v. Oracle is a significant decision for software developers, platform companies, and technology businesses that build on, interact with, or incorporate elements of third-party software systems.
The application of fair use doctrine to artificial intelligence — and specifically to the large-scale ingestion of copyrighted works to train machine learning models — is currently one of the most actively litigated and commercially significant copyright questions in the United States. Multiple lawsuits are pending against AI companies alleging that training on copyrighted text, images, code, and other content without authorization constitutes infringement not excused by fair use. AI companies have argued that training constitutes transformative use because the model does not reproduce the training data but uses it to learn patterns and statistical relationships. Copyright owners dispute this characterization, arguing that the output of AI models can closely mimic or substitute for the original works in their markets. As of the writing of this article, no definitive appellate or Supreme Court ruling on AI training and fair use has issued, and businesses in the AI space should monitor this area of law closely and engage qualified copyright counsel when assessing their exposure.
Advertising and Marketing
Businesses frequently encounter fair use issues in the context of advertising and marketing — both when creating their own materials and when assessing whether a competitor’s advertising crosses the line into infringement. Common scenarios include: using brief quotations from third-party reviews in advertising copy (generally permissible as factual quotation for informational purposes); reproducing a competitor’s product packaging for comparative advertising purposes (often defensible but context-dependent); using copyrighted music in social media content (highly risky and generally not excused by fair use given the well-developed licensing markets for music); and incorporating copyrighted photographs or artwork in marketing materials without a license (very unlikely to be fair use absent a strong transformative purpose).
The courts have been consistently skeptical of fair use claims in purely commercial advertising contexts where the primary purpose is to sell a product or service and where no commentary, criticism, or other speech interest is meaningfully served. The Andy Warhol Foundation decision reinforced this skepticism by making clear that aesthetic transformation alone is insufficient when the secondary use competes commercially with the original. Businesses that wish to use third-party creative works in advertising should obtain proper licenses rather than relying on fair use, which in this context is a high-risk defense.
Education and Training
Educational and training uses of copyrighted material are often assumed to be protected by fair use, but this assumption is more dangerous than most businesses realize. Section 107’s preamble lists “teaching” and “scholarship” among its illustrative examples, and courts do give weight to genuinely educational purposes. However, the educational character of a use does not override the other three factors, particularly the fourth. The reproduction of substantial portions of copyrighted textbooks, course packs, training materials, or professional reference works for internal corporate training programs — even where no charge is made to employees — can constitute infringement if it displaces licensing revenue that would otherwise flow to the copyright owner. Many publishers offer straightforward licensing programs specifically for corporate and educational use, and obtaining a license for commercially published training materials is far safer than attempting to rely on fair use.
Practical Risk Assessment for Businesses
Evaluating a Potential Fair Use Claim
When a business is considering using a copyrighted work without permission — or when it has been accused of infringement and is evaluating whether fair use is available as a defense — a structured analysis of all four statutory factors is essential. The analysis should be conducted honestly, giving genuine weight to factors that point against fair use as well as those that favor it.
Start with the purpose of the use. Is it transformative in the meaningful sense articulated by the Supreme Court in Campbell and Warhol — does it serve a genuinely different function or purpose from the original? Is it commercial? If it is commercial and not clearly transformative for a distinct purpose, the first factor likely weighs against you. Then assess the nature of the original: is it a highly creative work or a factual one? Published or unpublished? Consider the quantity and quality of what was taken: did you use only what was necessary for the stated purpose, and did you avoid taking the most distinctive or commercially valuable portions of the original? Finally, and most importantly, assess the market impact: does your use compete with the original or substitute for it in any market — including the licensing market — in which the copyright owner operates or might reasonably be expected to operate?
Courts also consider whether a license was available and, if so, whether the defendant sought one. A defendant who made no effort to seek a license when one was readily obtainable, and who then claims fair use after being sued, may find courts less receptive than one who had a principled, contemporaneous basis for concluding that the use was permissible. Documenting the basis for a fair use assessment — in writing, at the time of the decision — can be meaningful evidence of good faith.
When Fair Use Is and Is Not a Safe Strategy
Fair use is most reliably available when: the use is non-commercial or serves a strong public interest (commentary, criticism, parody, news, research); the amount taken is genuinely limited to what is necessary for the stated purpose; the original work is factual or informational rather than purely creative; and the use does not substitute for the original or for licensed derivatives in any recognizable market. In these circumstances, a well-documented fair use position is a reasonable basis on which to proceed without a license.
Fair use is a high-risk strategy when: the use is primarily commercial and serves no purpose meaningfully distinct from the original; the use involves substantial reproduction of a creative work; licensing of the original is readily available at reasonable cost; or the use clearly substitutes for the original in its primary or secondary markets. In these circumstances, obtaining a license is nearly always the wiser and more cost-effective approach — the cost of licensing is routinely far lower than the cost of defending a copyright infringement lawsuit, let alone the cost of an adverse judgment.
Businesses should also be aware that fair use is a defense, not a right, and that asserting it requires litigation to resolve if the copyright owner disagrees. The cost of defending even a meritorious fair use claim through to a final judgment can easily reach hundreds of thousands of dollars. For uses that carry meaningful fair use risk and where the commercial stakes are modest, the pragmatic calculation often favors obtaining a license regardless of the ultimate legal merits.
The Importance of Qualified Legal Counsel
Fair use analysis is fact-specific, contextual, and genuinely uncertain even for experienced practitioners. The Supreme Court has repeatedly resisted calls to establish bright-line rules, and the outcome of fair use cases often turns on subtle distinctions in how courts characterize the purpose and market impact of a secondary use. No legal guide, including this article, can substitute for individualized advice from a qualified copyright attorney who can assess the specific facts of your situation, review the works at issue, analyze the relevant case law in the applicable jurisdiction, and give you a considered judgment about the risks involved.
Fair Use and Artificial Intelligence: The Emerging Frontier
No discussion of fair use for US businesses would be complete without addressing the rapidly evolving legal questions surrounding artificial intelligence and copyright. As noted above, multiple significant lawsuits are currently pending that challenge the use of copyrighted works to train AI models. These cases — involving visual artists, novelists, journalists, musicians, software developers, and major media companies as plaintiffs — will likely produce appellate decisions within the next few years that substantially reshape the fair use landscape for technology companies.
At the heart of the AI fair use debate are several contested questions. First, does training a machine learning model on copyrighted works constitute “copying” in the legally relevant sense? AI training typically involves creating copies of training data, at least temporarily, which most courts would treat as reproduction triggering the Copyright Act. Second, is that copying transformative? AI developers argue that using works to train a model is fundamentally different from reproducing them for human consumption, and that the model does not “contain” the training works in any direct sense. Third, what is the relevant market for purposes of the fourth factor? If copyright owners develop licensing programs for AI training — as some are doing — the unauthorized use of their works for training may be found to harm a market that the owners are actively developing.
Businesses that develop, deploy, or rely on AI-generated content should stay closely informed about developments in this area. The legal landscape is changing rapidly, and a fair use analysis that appeared reasonable in 2023 may be significantly less defensible following appellate decisions expected in coming years. Engaging proactively with copyright counsel, evaluating licensing options where available, and building institutional knowledge about the provenance and licensing status of training data are prudent steps for any business with significant exposure in this area.
Conclusion
The fair use doctrine is one of the most nuanced, consequential, and genuinely uncertain areas of US intellectual property law. It reflects the law’s attempt to balance two sets of legitimate interests: the rights of creators and copyright owners to control the use of their works and earn a return on their creative investment, and the rights of the public — including subsequent creators, critics, journalists, researchers, and technology developers — to engage with, comment upon, and build upon the cultural and informational works that populate our shared environment.
For businesses, the doctrine is both an opportunity and a source of risk. It permits activities that copyright law would otherwise prohibit, enabling commentary, criticism, parody, news reporting, research, and certain transformative uses that enrich commerce and culture. But it does not provide a free pass for commercial exploitation of copyrighted works, and the boundaries between protected and infringing uses are drawn by courts case by case, applying a flexible four-factor test that resists reduction to simple rules.
The most important practical takeaway for US businesses is this: fair use is not a substitute for proper licensing when licensing is readily available and commercially reasonable. It is a defense for uses that serve genuine social and communicative functions that copyright law’s own built-in limitations — the idea/expression dichotomy, the originality requirement, the term of protection — do not fully address. Treat fair use as what it is: a context-specific, litigation-tested affirmative defense whose availability depends entirely on the specific facts of each situation. When significant amounts of third-party creative work are involved, when the use is primarily commercial, or when the stakes are high, the only prudent course is to obtain legal advice before proceeding. The cost of getting it right at the outset is invariably lower than the cost of getting it wrong.
This article is provided for general informational and educational purposes only and does not constitute legal advice.
Reading this article does not create an attorney-client relationship. Please consult a qualified attorney regarding your specific legal situation.