Summary. The doctrine that decides whether copying is infringement or freedom.

The most important sentence in copyright law is a hedge

Copyright gives an author a set of exclusive rights, listed in 17 U.S.C. § 106: reproduction, preparation of derivative works, distribution, public performance, public display, and digital audio transmission. Those rights are broad. Taken literally, they would make it unlawful to quote a paragraph in a book review, to include a poster in the background of a film, to record a television program, or to build a search engine.

They do not, because of 17 U.S.C. § 107:

"Notwithstanding the provisions of section 106 . . . , the fair use of a copyrighted work . . . for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright."

The statute then lists four factors to be considered "in determining whether the use made of a work in any particular case is a fair use":

  1. the purpose and character of the use, including whether it is of a commercial nature or is for nonprofit educational purposes;
  2. the nature of the copyrighted work;
  3. the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
  4. the effect of the use upon the potential market for or value of the copyrighted work.

Two things about this text drive everything that follows.

It is a standard, not a rule. Congress deliberately declined to define fair use, instead codifying in 1976 what courts had been doing since Justice Story's opinion in Folsom v. Marsh, 9 F. Cas. 342 (C.C.D. Mass. 1841). The House Report said the intention was "to restate the present judicial doctrine . . . not to change, narrow, or enlarge it in any way."

The factors are not a scorecard. Courts do not count factors and declare a winner. They weigh them together, in light of the purposes of copyright, and the weights shift depending on the context. A use can lose three factors and win.

The result is a doctrine that is genuinely hard to predict — which is a feature, not a bug, of a rule designed to accommodate uses nobody has thought of yet. It is also the reason fair use opinions are expensive and hedged.

Factor one: purpose and character

This factor asks what the second user did with the work, and it has become the center of gravity of the whole analysis.

The rise of "transformative use"

The modern era begins with Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569 (1994), the 2 Live Crew "Pretty Woman" case. The Court, borrowing a phrase from Judge Pierre Leval, asked:

"whether the new work merely supersedes the objects of the original creation, or instead adds something new, with a further purpose or different character, altering the first with new expression, meaning, or message; it asks, in other words, whether and to what extent the new work is transformative."

Campbell also made two moves that lasted:

  • It rejected a presumption against commercial uses. Commerciality is one element of factor one, not a trump. Most of the uses § 107 names — news reporting, criticism, comment — are ordinarily commercial.
  • It distinguished parody from satire. Parody targets the original work and therefore needs to borrow from it. Satire uses the original to comment on something else and therefore has less justification for taking. This distinction sounds academic and decides cases.

For the next quarter century, "transformative" did enormous work. Courts found transformation in thumbnails for image search, in full-text scanning for a searchable book database, in appropriation art, in reference guides, and in software reimplementation. The word became, in Judge Leval's later worried phrase, close to a talisman.

Google v. Oracle

Google LLC v. Oracle America, Inc., 593 U.S. 1 (2021) applied the framework to software. Google had copied roughly 11,500 lines of declaring code from the Java SE API to let Java programmers use their existing skills on Android.

The Court assumed without deciding that the code was copyrightable and held the use fair. Its reasoning is distinctly software-shaped:

  • Factor two favored Google because declaring code is "further than are most computer programs" from the core of copyright — its value comes substantially from the investment of programmers who learned it, not from the creativity of its author.
  • Factor one favored Google because reimplementing an interface to allow programmers to work in a new environment is transformative — it creates a new platform rather than substituting for the old.
  • Factor three favored Google because the 11,500 lines were a small fraction of the whole and were taken because they were needed for interoperability, not for their expressive quality.
  • Factor four favored Google because Java SE's market was desktop computing, Android's was smartphones, and Sun had not succeeded in the mobile market. The Court also emphasized the "public benefits" of allowing programmers to reuse skills, and the risk that enforcement would let the interface holder lock in a workforce.

Google v. Oracle is best read as a case about interfaces and interoperability rather than a general expansion of fair use. But it demonstrated the Court's willingness to treat factor four as an inquiry into real-world market effects rather than a formal question about lost licensing revenue.

Warhol and the discipline of "same purpose"

Then came Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, 598 U.S. 508 (2023), which did not overrule anything but substantially reoriented factor one.

The facts matter. Lynn Goldsmith photographed Prince in 1981. In 1984 Vanity Fair licensed the photograph as an "artist reference" for an illustration; Andy Warhol made a silkscreen from it, which ran with the article. Warhol also made fifteen additional works from the photograph that were not licensed. In 2016, after Prince's death, the Warhol Foundation licensed one of those — Orange Prince — to Condé Nast for a magazine cover.

Goldsmith sued. The question presented was narrow: whether that 2016 licensing of Orange Prince to a magazine was a fair use.

The Court, in an opinion by Justice Sotomayor, held it was not. The reasoning:

"The first fair use factor . . . considers the reasons for, and nature of, the copier's use of an original work. The 'central' question it asks is 'whether the new work merely supersedes the objects of the original creation . . . or instead adds something new . . . .' In that way, the first factor relates to the problem of substitution — copyright's concern with the original work's value."

And the key holding:

"[T]he first factor . . . considers whether the use of a copyrighted work has a further purpose or different character, which is a matter of degree, and the degree of difference must be balanced against the commercial nature of the use. If an original work and secondary use share the same or highly similar purposes, and the secondary use is of a commercial nature, the first factor is likely to weigh against fair use, absent some other justification for copying."

Both Goldsmith's photograph and Orange Prince were used for the same purpose: to illustrate a magazine article about Prince. That shared purpose, combined with commerciality, defeated factor one.

What Warhol changed:

  • The unit of analysis is the use, not the work. The question is not "is Orange Prince a transformative artwork?" — the Court accepted that Warhol added new expression and meaning. The question is "was this particular use of it transformative relative to this particular use of the original?" The Court was explicit that its holding did not address whether Warhol's creation of the series was infringing, or whether display in a museum would be fair.
  • New meaning is not enough. Adding expression, meaning, or message is relevant but not sufficient. A use that adds meaning while serving the same market function is a substitute.
  • Justification matters. The Court repeatedly asked whether the copier had a reason to use this work rather than another — the parody/satire distinction from Campbell, generalized. A user who needed the original to make the point has a justification; one who used it because it was convenient does not.

What Warhol did not change:

  • Transformative use remains a real and important concept. Campbell and Google were distinguished, not overruled.
  • Non-commercial and clearly different-purpose uses — criticism, commentary, scholarship, search indexing — are not affected.
  • The four-factor analysis remains holistic.

The dissent by Justice Kagan, joined by the Chief Justice, argued vigorously that the majority's approach would chill creative reuse. Whether it has is an empirical question the lower courts are still working out. What is clear is that the phrase "it's transformative" is no longer a complete argument.

Factor two: the nature of the copyrighted work

This factor asks where the work sits relative to the core of copyright's concern. Two axes matter:

Creative versus factual. Highly creative works — novels, songs, photographs, films — get more protection. Factual and functional works — directories, technical manuals, news accounts, software interfaces — get less, because copyright does not protect facts or ideas, only expression.

Published versus unpublished. Copying from an unpublished work weighs against fair use, because the author's right of first publication is significant. Harper & Row Publishers, Inc. v. Nation Enterprises, 471 U.S. 539 (1985) — The Nation's scoop of President Ford's memoirs — placed heavy emphasis on this. Congress later amended § 107 to add that "[t]he fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made upon consideration of all the above factors," softening but not eliminating the point.

Factor two rarely decides a case on its own. It did real work in Google v. Oracle, where the functional nature of declaring code shaped the whole analysis.

Factor three: amount and substantiality

Both quantitative and qualitative. Copying a small portion helps; copying a large portion hurts; but copying the "heart" of a work hurts even if it is short. Harper & Row involved roughly 300 words out of a 200,000-word manuscript — but they were the words about the Nixon pardon, the most newsworthy passage in the book.

Two principles matter more than the raw percentage:

The amount must be reasonable in relation to the purpose. Campbell framed factor three as asking whether the quantity and value of the materials used are reasonable in relation to the purpose of the copying. A parodist needs enough to conjure up the original. A search engine needs the whole work to index it. A critic needs enough to make the criticism intelligible.

Copying the whole work is not fatal. Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417 (1984) held that time-shifting a whole broadcast program for home viewing was fair use. Full-text scanning for search has been held fair. What matters is whether the full taking is necessary for the transformative purpose and whether the result substitutes for the original.

Factor four: market effect

Harper & Row called this "undoubtedly the single most important element of fair use," and while Campbell rejected any rigid hierarchy, factor four remains the anchor — and after Warhol, factor one and factor four are explicitly linked through the concept of substitution.

The inquiry has two parts:

Harm to the market for the original. Does the use substitute for purchases of the original? A book review does not; a pirated copy does.

Harm to the market for derivative works. Does the use occupy a licensing market the copyright owner is entitled to exploit? This is where the analysis gets circular, and courts know it: any use could in principle be licensed, so treating every unlicensed use as market harm would swallow the doctrine. Courts therefore ask whether the market is one that copyright owners traditionally, reasonably, or likely would develop — excluding markets that exist only because of the challenged use itself.

The Warhol synthesis. Warhol made the relationship explicit: because Goldsmith licensed photographs to magazines and the Foundation licensed Orange Prince to a magazine, the two were competing in the same market. The shared purpose under factor one was the market substitution under factor four.

Aggregate effects count. Campbell noted that the inquiry considers not only the challenged use but "whether unrestricted and widespread conduct of the sort engaged in by the defendant . . . would result in a substantially adverse impact on the potential market."

Where fair use is actually litigated

Parody and commentary

The most doctrinally comfortable territory. A work that comments on the original has a justification for borrowing from it. The recurring failure is the satire problem: using a famous work to make a point about something else. Courts also distinguish commentary from mere evocation — a work that borrows an aesthetic without saying anything about the source is on thin ground after Warhol.

News reporting

Section 107 lists news reporting, but Harper & Row shows the limits. Reporting about a work, quoting what is necessary, is well protected. Reproducing the work as the news — scooping the publication itself — is not.

Education and scholarship

Classroom copying, course packs, and library reserves have generated substantial litigation. Nonprofit educational purpose helps under factor one, but factor four remains live where a licensing market exists — the permissions market for academic excerpts is well developed, and courts have held that copying that displaces it is not fair.

Software and interoperability

Google v. Oracle is the touchstone. Reverse engineering to achieve interoperability has a long line of favorable authority. Copying code for its expressive content does not.

Search, indexing, and text mining

The search cases — thumbnails, full-text book search, plagiarism-detection databases — share a structure: the copying is complete, the output is not a substitute, and the purpose is to provide information about works rather than the works themselves. Where a system's output reproduces expression rather than merely locating it, the analysis changes.

Machine learning

The newest and most consequential frontier. The core arguments track the search cases: developers argue that training extracts statistical relationships rather than expression, that the purpose differs entirely from the original, and that models do not substitute for the works. Rights holders argue that training reproduces works wholesale, that outputs can compete directly with the originals, and that a licensing market for training data exists and is being displaced.

Warhol matters here in a specific way. If a model is used to generate outputs that serve the same purpose as the works it trained on — illustrations competing with the illustrators, articles competing with the articles — the shared-purpose analysis cuts against fair use even if the training process is technically transformative. If it is used for research, analysis, or purposes unrelated to the original market, the analysis is friendlier. The distinction between training and output is doing a great deal of work in current litigation, and courts have begun to treat them as separate questions.

Related doctrines matter too. 17 U.S.C. § 1202 prohibits removing or altering copyright management information, and claims under it are appearing in training-data cases independent of the fair use question.

Three worked examples

Fair use is best understood by running it. Here are three uses that look similar and come out differently.

Example one: the documentary

Beatriz Ocampo is making a documentary about the collapse of a regional airline. She wants to include forty seconds of a 1997 television commercial the airline ran — a cheerful jingle over footage of smiling flight attendants — immediately before an interview with a former mechanic describing maintenance shortcuts.

Factor one. The purpose is criticism and commentary: the juxtaposition makes an argument about the gap between the airline's public image and its operations. This is a different purpose from the commercial's original purpose, which was to sell tickets. Beatriz needs this commercial, not a generic one, because the point is about this airline's specific self-presentation. That is the Campbell justification. The documentary is commercial, which cuts the other way, but under Campbell commerciality is not dispositive and most criticism is commercial.

Factor two. The commercial is creative, which cuts against Beatriz — but the factor is weak here because the use is commentary rather than exploitation of the creativity.

Factor three. Forty seconds of a sixty-second commercial is a large proportion. But the amount is reasonable in relation to the purpose: a five-second clip would not establish the tone she is critiquing. Courts have accepted substantial takings in documentary criticism.

Factor four. There is no market in which a forty-second excerpt inside a critical documentary substitutes for the commercial. The airline is defunct; the commercial has no ongoing licensing market except for exactly this kind of use — and courts exclude markets that exist only because of the challenged use.

Conclusion: strong fair use. Beatriz's counsel should still document the analysis, because errors-and-omissions insurers require it.

Example two: the stock photograph on a blog

Torsten Aalborg runs a marketing consultancy and writes a blog post about brand storytelling. He finds a striking photograph of a lighthouse on a photographer's portfolio site and uses it as the header image, with a credit line and a link.

Factor one. The purpose is decorative. The photograph illustrates nothing about itself; it makes the post look professional. That is the same purpose the photograph was created for and is licensed for. Under Warhol, shared purpose plus commercial use weighs heavily against fair use. The credit line is irrelevant. Torsten had no reason to need this photograph; any lighthouse image would have served.

Factor two. Creative photograph. Against.

Factor three. The entire work. Against.

Factor four. Directly against. Stock photography licensing is exactly the market being displaced, and it is a traditional and well-developed one.

Conclusion: not fair use, and not close. This is the single most common copyright mistake made by businesses, and it generates a large volume of demand letters. Statutory damages under 17 U.S.C. § 504 are available if the photograph was timely registered under 17 U.S.C. § 412.

Example three: the search index

Corvid Labs builds a tool that ingests every publicly available municipal building code in the United States, indexes the text, and lets users search across jurisdictions. Results show the searched phrase in context — about two sentences — with a citation and a link to the official source.

Factor one. The purpose is entirely different from the codes' purpose. The codes exist to regulate construction; the index exists to help people find provisions across jurisdictions. This is the classic search-engine posture, and it is transformative in the sense Warhol preserved: the tool tells you about the works rather than substituting for them.

Factor two. Building codes are functional and largely factual, sitting far from copyright's core. Where a code has been enacted into law, an additional and stronger argument applies — the government edicts doctrine holds that the law itself is uncopyrightable.

Factor three. Complete copying, necessary for indexing. The output is two sentences.

Factor four. The two-sentence snippet does not substitute for the code. A user who needs the provision must go to the source.

Conclusion: strong fair use for the index. The analysis would change if Corvid displayed whole sections, or if it built a product that let users skip the official source entirely.

How the factors actually interact

Reading the cases, a few patterns emerge that the statute does not state.

Factors one and four are joined. Warhol made explicit what had been implicit: whether the use shares the original's purpose (factor one) largely determines whether it substitutes in the original's market (factor four). A use that serves a genuinely different function rarely competes; a use that serves the same function almost always does.

Factors two and three are usually subordinate. They rarely decide cases alone. Their function is to modulate: factor two tells you how much protection the work deserves, factor three tells you whether the taking was proportionate to the justification established under factor one.

Justification is the hidden fifth factor. Across the cases, the question that recurs is: did the user need this particular work? A parodist does. A critic does. A programmer seeking interoperability does. A blogger who wanted a nice header image does not. This is not in the statute, but it explains outcomes better than any single factor.

Bad faith matters less than it used to. Harper & Row mentioned the purloined manuscript. Campbell cautioned against making bad faith dispositive, and Google v. Oracle declined to treat Google's knowledge that it lacked a license as significant. Bad faith is still a bad fact, but it is not a separate element.

Advising on fair use

Clients want yes or no. The doctrine gives degrees of risk. Practical guidance for both sides of that conversation:

Do the analysis in writing. A contemporaneous memorandum applying the four factors is the best evidence that the use was considered in good faith. It matters for willfulness under 17 U.S.C. § 504(c)(2), which permits reduced statutory damages where the infringer had reasonable grounds to believe the use was fair, and it can eliminate statutory damages entirely for certain nonprofit educational and library uses.

Frame the answer as risk, not permission. "This is a strong fair use position" and "this is a weak one" are honest. "This is fair use" usually is not, unless the case is genuinely clear.

Identify the market you are being accused of displacing. Most fair use fights are factor four fights. If you can articulate why the use does not substitute — and better, why no reasonable licensing market exists for this kind of use — you have most of the argument.

Ask whether a license is cheap. A great many fair use questions are asked about uses that could be licensed for a few hundred dollars. When the license is cheap and available, the legal analysis is often beside the point. Reserve the fair use argument for uses that cannot be licensed: criticism the owner would refuse to permit, research at scale, interoperability, and commentary.

Document the alternatives you considered. Evidence that you looked for a licensable alternative and found none supports the justification analysis. Evidence that you did not look supports the opposite.

Insurance and platform policies matter. Errors-and-omissions insurers for film and publishing require fair use opinions with specified elements. Platform content policies impose their own standards that are frequently stricter than copyright law. A use may be legally fair and still get taken down under 17 U.S.C. § 512, with the counter-notice process as the remedy.

What fair use is not

It is not a defense you get for free. Fair use is an affirmative defense. The user bears the burden.

It is not the same as "I gave credit." Attribution is good practice and is irrelevant to the four factors, except insofar as it bears on the character of the use.

It is not a percentage. There is no 10% rule, no thirty-second rule for music, no five-line rule for poetry. Those numbers come from guidelines negotiated among interest groups, not from law.

It does not protect against a license breach. Fair use is a limit on copyright. If you agreed by contract not to copy, fair use does not excuse the breach. Terms of service, database licenses, and platform agreements routinely restrict uses that copyright itself would permit.

It does not resolve other claims. A use that is fair under copyright may still violate the right of publicity, trademark law, or a confidentiality obligation.

Fair use in litigation

The doctrine looks different from inside a case than it does in a clearance memorandum.

It is an affirmative defense, and the burden is real. The defendant must plead and prove it. That allocation shapes discovery: the copyright owner establishes ownership and copying, and then the defendant must build a record on all four factors, including the negative — that no market was harmed.

Summary judgment is common but not assured. Where the facts about the use are undisputed, courts frequently resolve fair use as a matter of law. Where the purpose of the use, the nature of the market, or the extent of substitution is genuinely contested, the case goes to trial. Google v. Oracle clarified the appellate posture: the ultimate fair use determination is a legal question reviewed de novo, with subsidiary factual findings preserved.

Factor four requires evidence, not assertion. The most common failure in fair use litigation is a defendant who argues market harm is absent without putting in any evidence about the market. Expert testimony about licensing practices, survey evidence about substitution, and testimony from market participants all help. So does evidence that the plaintiff has never licensed for this kind of use and has no plans to.

Discovery targets. A defendant should seek the plaintiff's licensing history for comparable uses, evidence of any licensing market for the specific type of use at issue, revenue data showing whether the challenged use correlated with any decline, and internal documents discussing the challenged use. A plaintiff should seek the defendant's internal analysis of the use, evidence about alternatives considered, revenue attributable to the use, and communications showing awareness of the licensing option.

Remedies exposure shapes settlement. Statutory damages under 17 U.S.C. § 504 range from $750 to $30,000 per work, rising to $150,000 for willful infringement and falling to $200 for innocent infringement. Because the range is per work, a case involving hundreds of works has a theoretical exposure that dwarfs any actual harm — which is why fair use defendants in mass-copying cases face settlement pressure disproportionate to the merits. Attorney fees under § 505 are discretionary and available to prevailing defendants.

The timing rules bite. Registration is generally a prerequisite to suit under 17 U.S.C. § 411 after Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC, 586 U.S. 296 (2019), and 17 U.S.C. § 412 bars statutory damages and fees for infringement commencing before registration (with a three-month grace period for published works). Unicolors, Inc. v. H&M Hennes & Mauritz, L.P., 595 U.S. 178 (2022) held that a registration is not invalidated by an inaccuracy the applicant made through a mistake of law, which narrowed a common defense. And Petrella v. Metro-Goldwyn-Mayer, Inc., 572 U.S. 663 (2014) held that laches cannot bar a claim brought within the three-year limitations period.

The international dimension

Fair use is an American doctrine. Most of the world uses fair dealing or a closed list of specific exceptions, and the difference is structural rather than semantic.

Fair dealing jurisdictions — the United Kingdom, Canada, Australia, India, and others in the common law tradition — permit unlicensed use only for enumerated purposes: research and private study, criticism and review, news reporting, parody in some jurisdictions, quotation. If the use does not fall within a listed purpose, the analysis ends; there is no open-ended balancing.

European Union law operates through an exhaustive list of optional exceptions in the InfoSoc Directive, implemented differently by each member state, plus specific provisions for text and data mining that permit research use and allow rightsholders to opt out of commercial mining.

Practical consequences for multinational uses:

  • A use that is fair in the United States may infringe elsewhere. Global product launches require country-by-country analysis, not a single memorandum.
  • Contracts frequently allocate this risk through representations and indemnities. Read them carefully: a warranty of non-infringement "in any jurisdiction" is far broader than it looks.
  • Moral rights — attribution and integrity — exist robustly in many jurisdictions and have no fair use analogue. A use that is permissible as to economic rights may still violate the author's moral rights in France or Germany.
  • For online services, the location of servers, users, and the targeted market all matter, and the answers differ across the copyright, jurisdictional, and regulatory analyses.

A short history, because it explains the shape of the doctrine

Fair use did not come from Congress. It came from Joseph Story, sitting as a circuit justice in 1841, deciding whether Charles Upham could publish a biography of George Washington that reproduced 353 pages of letters from Jared Sparks's twelve-volume edition.

Story's formulation in Folsom v. Marsh is recognizably the modern test:

"[W]e must often . . . look to the nature and objects of the selections made, the quantity and value of the materials used, and the degree in which the use may prejudice the sale, or diminish the profits, or supersede the objects, of the original work."

Purpose, amount, market effect. Three of the four factors, in one sentence, 135 years before the statute.

For the next century and a quarter, fair use developed entirely in the courts, which is why it retains the texture of common law reasoning inside a codified statute. When Congress finally codified it in the 1976 Act, it deliberately declined to define it, and the legislative history is explicit that the courts should continue to develop it case by case.

Two consequences follow, and they are worth stating plainly because they frustrate clients.

First, precedent works differently here. Because the analysis is holistic and fact-specific, a prior case is rarely dispositive. A lawyer citing Campbell for the proposition that a use is transformative is making an analogy, not applying a rule. This is why fair use opinions read as they do.

Second, the doctrine is designed to be applied to facts nobody anticipated. Photocopiers, videocassette recorders, search engines, and machine learning were all confronted with a test written before any of them existed, and in each case the test proved capable of producing a defensible answer. That adaptability is bought with unpredictability, and Congress made that trade deliberately.

Understanding this history helps in a practical way: it tells you that arguments grounded in the purposes of copyright — encouraging creation, avoiding lock-up of facts and ideas, preventing substitution — carry weight, while arguments grounded in mechanical application of factor language do not.

Frequently asked questions

Does fair use apply to the DMCA takedown process? Yes, and it is required. A copyright owner sending a notice under 17 U.S.C. § 512 must consider whether the use is fair before certifying a good-faith belief that it is unauthorized. A user whose material is removed may file a counter-notice, after which the material is restored unless the owner sues. In practice, platform policies frequently exceed what copyright requires, and the takedown system resolves far more disputes than the courts do.

Can a use become unfair over time? Effectively, yes. Factor four asks about the market as it exists, and licensing markets develop. A use that displaced no market in 2010 may displace a well-established one in 2026. This is one reason fair use opinions should carry a date and a recommendation to revisit.

Does Warhol mean transformative use is dead? No. It means transformation is measured against the specific use, and that adding new meaning does not overcome a shared commercial purpose. Uses with genuinely different purposes — criticism, research, indexing — are unaffected.

Is non-commercial use automatically fair? No, but it helps under factor one and weakens the market-harm showing under factor four.

Can I use a whole work? Sometimes. Sony and the search cases show that complete copying can be fair where it is necessary to a different purpose and does not substitute.

Does registration matter to fair use? Not to the analysis, but it matters enormously to remedies. Under 17 U.S.C. § 411 and Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC, 586 U.S. 296 (2019), a plaintiff generally must have a registration in hand to sue, and 17 U.S.C. § 412 conditions statutory damages and fees on timely registration.

Who decides fair use — judge or jury? Fair use is a mixed question. Google v. Oracle treated the ultimate question as legal, reviewed de novo, with subsidiary factual findings for the jury. Many cases are resolved on summary judgment.

Does a disclaimer help? Rarely. A disclaimer does not change the four factors, though it can matter for trademark and false endorsement claims.

What about the "de minimis" doctrine? Copying so trivial that no substantial similarity exists is not infringement at all, and never reaches fair use. The two are often confused. Courts split on how de minimis operates for digital sampling.

Related documents