Federal Jurisdiction Over Copyright: Section 1338(a) Exclusivity, the Registration Prerequisite, Accrual, Preemption, and Fee-Shifting

Copyright is one of the two fields where Congress gave the federal courts exclusive jurisdiction and then left the boundary undefined. Section 1338(a) says state courts may not hear a copyright claim; it does not say what a copyright claim is. The result is that a large share of copyright motion practice is not about substantive copyright law at all. It is about whether the pleading in front of the court is an infringement claim or a contract claim, whether a registration certificate existed on the day the complaint was filed, when the three-year clock started, and whether the plaintiff owns an exclusive right at all.

Those threshold questions are unusually consequential in this field because the sanctions for getting them wrong are asymmetric and often terminal. A plaintiff who sues before the Register of Copyrights acts on the application has filed a complaint that will be dismissed, and the delay may cost the statutory damages and fee award that made the case worth bringing. A plaintiff who characterizes an ownership dispute as infringement may be sent to state court after two years of federal discovery. A defendant who treats a registration inaccuracy as a silver bullet is, after 2022, usually wrong.

This guide covers the jurisdictional grant and its boundary with state law, the registration prerequisite and the two Supreme Court decisions that moved it twice in a decade, the limitations and accrual rules as they now stand after Warner Chappell, remedies and fee-shifting, express preemption and the copyright/trademark boundary, state sovereign immunity, and the substantive doctrines — originality, work for hire, separability, fair use, secondary liability — that determine who may sue and what the claim is worth.

Contents

• Exclusive jurisdiction under § 1338(a)
• The contract/infringement boundary
• Venue under § 1400(a)
• The registration prerequisite
• Inaccurate registrations and § 411(b)
• Limitations and accrual under § 507(b)
• Remedies and fee-shifting
• Preemption under § 301
• Dastar and the copyright/trademark boundary
• Sovereign immunity and government works
• Substantive doctrines that decide who may sue
• Fair use
• Secondary liability and technology
• Standing, joinder, and supplemental jurisdiction
• Practice checklist
• Frequently asked questions

Exclusive jurisdiction under § 1338(a)

28 U.S.C. § 1338(a) gives the district courts original jurisdiction over "any civil action arising under any Act of Congress relating to patents, plant variety protection, copyrights and trademarks," and makes that jurisdiction exclusive of the state courts as to patents, plant variety protection, and copyrights. Trademark is the odd one out: § 1338(a) reaches Lanham Act claims but does not make them exclusive, which is why trademark cases routinely proceed in state court and copyright cases do not.

Two features of the grant matter immediately. First, there is no amount-in-controversy requirement and no diversity requirement — a copyright claim between two citizens of the same state for $2,000 in statutory damages belongs in federal court and nowhere else. Second, exclusivity runs to the claim, not the subject matter. A contract dispute about a copyright is not a copyright claim, and § 1338(a) does not pull it into federal court. That distinction generates most of the litigation under this statute.

The "arising under" language in § 1338(a) is read in parallel with the same phrase in 28 U.S.C. § 1331, and the Supreme Court has repeatedly said so. Christianson v. Colt Industries Operating Corp. held that § 1338(a) jurisdiction exists only where the plaintiff's right to relief "necessarily depends on resolution of a substantial question of federal patent law, in that patent law is a necessary element of one of the well-pleaded claims" — a formulation lifted directly from the § 1331 well-pleaded-complaint cases and applied without modification to copyright. Holmes Group, Inc. v. Vornado Air Circulation Systems, Inc. then confirmed that a counterclaim cannot create § 1338 jurisdiction, because the well-pleaded complaint rule looks to the plaintiff's claims alone. Congress partially overrode Holmes Group for patents in 28 U.S.C. § 1454, which permits removal on a patent counterclaim; there is no copyright analogue, so a copyright counterclaim in a state contract action does not support removal and the state court must resolve it or the parties must file separately.

Gunn v. Minton supplies the modern framework and is worth reading closely by anyone litigating the § 1338 boundary, because it reasons from § 1338 cases to reach a § 1331 conclusion. The Court restated Grable & Sons Metal Products, Inc. v. Darue Engineering & Manufacturing as four conjunctive requirements — the federal issue must be necessarily raised, actually disputed, substantial to the federal system as a whole, and resolvable without disrupting the federal-state balance — and held that a legal-malpractice claim requiring a state court to decide a hypothetical patent question satisfied none of the last two. Gunn made explicit what practitioners had long assumed: state courts are competent to resolve embedded federal intellectual-property questions, and § 1338(a) exclusivity does not strip them of that competence. The full treatment of the Grable/Gunn analysis is in the federal subject-matter jurisdiction guide; the point here is that it governs the copyright boundary as squarely as it governs everything else.

The contract/infringement boundary

The framework the circuits actually apply comes from Judge Friendly's opinion in T.B. Harms Co. v. Eliscu, a 1964 Second Circuit decision that the Supreme Court has never reviewed and that every circuit has adopted in substance. The test asks whether the complaint:

1. seeks a remedy expressly granted by the Copyright Act — an injunction under 17 U.S.C. § 502, damages or profits under § 504, impoundment under § 503;
2. asserts a claim requiring construction of the Act — the scope of § 106 exclusive rights, the work-for-hire definitions in § 101, the termination provisions in §§ 203 and 304(c); or
3. presents a distinctive policy of the Act requiring federal principles of construction.

If none of the three is satisfied, the claim does not arise under the Act, whatever the subject matter. The practical shorthand every district court uses: does the complaint allege infringement, or does it allege breach? A plaintiff who says "the defendant copied my work without authorization" pleads infringement. A plaintiff who says "the defendant exceeded the scope of the license, or failed to pay the royalty, or refuses to acknowledge that I am a co-author under our agreement" is usually pleading breach, and the fact that a copyright is the thing being fought over does not change the answer.

The license-scope line is where the fights are. The standard analytical move — adopted from the Ninth Circuit's Sun Microsystems v. Microsoft line and followed broadly — distinguishes a license condition from a license covenant. Breach of a condition means the licensee's use falls outside the grant, so the use is unlicensed and infringing, and federal jurisdiction follows. Breach of an independent covenant leaves the license intact; the remedy is contractual. Whether a given clause is a condition or a covenant is a question of the governing state contract law, which is a genuine oddity: state law determines whether the federal court has jurisdiction. Draft accordingly if you are the licensor — a clause framed as a limitation on the scope of the grant preserves the infringement remedy in a way that a clause framed as a payment obligation does not.

A related and frequently mishandled posture is the ownership dispute. Where the parties dispute ownership under a contract — an assignment whose validity is contested, a partnership dissolution, an employment agreement's scope — the claim ordinarily belongs in state court, because resolution requires construing the instrument rather than the Act. Where the dispute turns on whether a work qualifies as a work made for hire under the § 101 definitions, or on who counts as an author of a joint work, the Act's provisions are essential and federal jurisdiction is proper. The line is not always obvious, and the Second Circuit's practice of asking whether the ownership question is "merely incidental" to an infringement claim — in which case the federal court resolves both — is the safest formulation to brief.

A defendant who wants the case out of federal court should also remember what the plaintiff cannot do. Under the well-pleaded complaint rule, as applied in Louisville & Nashville Railroad Co. v. Mottley and restated in the § 1338 context by Christianson, a plaintiff cannot manufacture § 1338 jurisdiction by anticipating a copyright defense to a contract claim. A licensee sued for breach cannot remove by asserting that its use was fair; a plaintiff suing on a contract cannot plead into federal court by predicting a copyright defense.

Venue under § 1400(a)

28 U.S.C. § 1400(a) governs venue in copyright and mask-work cases: such actions "may be instituted in the district in which the defendant or his agent resides or may be found." That language is substantially broader than the patent provision in § 1400(b), which after TC Heartland LLC v. Kraft Foods Group Brands LLC confines patent venue to the defendant's state of incorporation or a district where it has a regular and established place of business.

Two points follow. First, "may be found" is read to mean amenable to personal jurisdiction in the district, so copyright venue effectively collapses into the personal-jurisdiction inquiry — if the court can reach the defendant, venue lies. That gives a copyright plaintiff meaningfully wider forum choice than a patent plaintiff, and the general-jurisdiction and stream-of-commerce analysis in the personal jurisdiction over corporations guide does most of the work. Second, § 1400(a) is permissive rather than exclusive; the general venue statute in 28 U.S.C. § 1391 remains available, and a plaintiff may rely on whichever provision fits. Transfer under 28 U.S.C. § 1404 is the realistic defense response, not a venue dismissal.

The registration prerequisite

This is the single most consequential procedural issue in copyright litigation, and the doctrine moved twice in a decade. 17 U.S.C. § 411(a) provides that "no civil action for infringement of the copyright in any United States work shall be instituted until preregistration or registration of the copyright claim has been made in accordance with this title." 17 U.S.C. § 412 then conditions the two remedies that make most copyright cases economically viable — statutory damages under § 504(c) and attorney's fees under § 505 — on the timing of registration relative to the infringement.

Reed Elsevier: not jurisdictional

Reed Elsevier, Inc. v. Muchnick (2010) arose from a class settlement of claims by freelance authors over electronic republication of their articles. The class included owners of both registered and unregistered works; the Second Circuit held, on its own motion and over the objection of every party, that § 411(a) deprived the district court of jurisdiction to approve a settlement covering unregistered works.

The Supreme Court reversed. Section 411(a) "does not restrict a federal court's subject-matter jurisdiction"; it is a precondition to filing a claim — a claim-processing rule — and so does not deprive a court of authority to adjudicate claims involving unregistered works, at least where the parties do not press the point. The Court placed the holding squarely in the Arbaugh v. Y & H Corp. line: where Congress does not clearly rank a threshold requirement as jurisdictional, courts should treat it as an element or a filing precondition. The textual signals mattered — § 411(a) appears in a section about registration procedure rather than in a jurisdictional provision, it contains express exceptions, and § 1338(a) separately supplies jurisdiction over copyright actions without reference to registration.

What Reed Elsevier changed and what it did not:

• Changed. A registration objection can be waived or forfeited if not timely raised. A court need not police it on its own motion. A settlement, a consolidated class, or a consent judgment may reach unregistered works. An appellate court cannot vacate a judgment years later because a plaintiff's certificate was missing.
• Did not change. Section 411(a) is still a mandatory precondition when a defendant invokes it. A defendant who raises it in a Rule 12(b)(6) motion is entitled to dismissal of claims on unregistered works. The remedy-timing rules in § 412 were untouched.

The tactical lesson for defendants is the mirror image of the holding. Because the objection is now forfeitable, it must be raised in the first responsive pleading and pressed. Because it is not jurisdictional, a defendant cannot hold it in reserve as an appellate insurance policy.

Fourth Estate: registration means the Register acted

Reed Elsevier left open what "registration … has been made" means, and the circuits split badly. Under the application approach (Fifth and Ninth Circuits, and the majority of district courts nationwide), a plaintiff satisfied § 411(a) by submitting a complete application, deposit, and fee. Under the registration approach (Tenth and Eleventh Circuits), the Register of Copyrights had to have acted on the application first.

Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC (2019) resolved it unanimously for the registration approach. Registration "has been made" when the Register grants the registration — or refuses it, since § 411(a) permits suit after refusal so long as notice is served on the Register. The Court's reasoning was structural: § 411(a)'s second sentence allows an applicant to sue "if registration has been refused," which would be superfluous if mere application sufficed; § 410(d) sets the effective date of a registration at the date the complete application was received, which would be unnecessary if the application itself triggered suit rights; and the preregistration alternative in § 408(f) exists precisely for works especially vulnerable to pre-distribution infringement, an accommodation that would be pointless under the application approach. The Court acknowledged the administrative delay problem and answered it institutionally: the remedy for a slow Copyright Office is an appropriation, not a judicial rewrite of the statute.

The practical consequences are immediate and unforgiving:

• The wait. Processing times for ordinary electronic applications have run from several months to well over a year. During that window the copyright owner has a valid copyright — registration is not a condition of the copyright itself, only of the suit — and no enforceable claim in federal court.
• Special handling. The Copyright Office will expedite an application for a substantial additional fee where there is pending or prospective litigation, a customs matter, or a contract or publishing deadline. This is the standard answer to the Fourth Estate problem and should be initiated the day litigation becomes likely. Special handling has historically produced a certificate in a matter of days to a few weeks.
• Preregistration. Section 408(f) preregistration is available only for specified classes of works being prepared for commercial distribution — motion pictures, sound recordings, musical compositions, literary works being prepared in book form, computer programs, advertising photographs — and it triggers an obligation to complete registration within a set period after publication. It is a narrow tool for pre-release piracy, not a general substitute.
• Refusal is a ticket. If the Register refuses registration, the applicant may still sue under § 411(a) provided a copy of the complaint is served on the Register. A refusal is therefore not the end of the case, and a plaintiff whose application is rejected should not abandon the claim.
• Register the portfolio in advance. For any business with a continuing output of protectable works — software releases, photography, music catalogs, database compilations — routine prompt registration is the only reliable answer, and it is also what § 412 independently demands.

Section 412 and the remedies clock

Section 412 is the reason prompt registration matters even for a plaintiff willing to wait out the Office. No award of statutory damages or attorney's fees may be made for:

1. any infringement of an unpublished work commenced before the effective date of its registration; or
2. any infringement of a published work commenced after first publication and before the effective date of registration, unless registration was made within three months after first publication.

Three features of this provision cause repeated losses. First, the operative date is the date infringement commenced, not the date each infringing act occurred — so a defendant whose infringement began before registration generally escapes statutory damages and fees for the entire continuing course of conduct, including acts after registration. Second, the three-month grace period for published works runs from first publication and is unforgiving; a registration filed in month four cannot be backdated. Third, the effective date of registration under § 410(d) is the date the Copyright Office received a complete application, deposit, and fee — which means a long pendency at the Office does not push the § 412 date later, even though it does postpone the day suit may be filed under Fourth Estate. That asymmetry is the one piece of good news in this area: file the application early and the § 412 date is fixed, even if the certificate arrives much later.

A plaintiff barred from statutory damages by § 412 is left with actual damages and the infringer's profits under § 504(b), which require proof, expert work, and a causal nexus — and with no fee award to fund any of it. The economics of a great many copyright cases turn entirely on this section.

Inaccurate registrations and § 411(b)

A certificate in hand is not the end of the registration fight. Section 411(b), added by the PRO IP Act of 2008, provides that a certificate satisfies § 411(a) "regardless of whether the certificate contains any inaccurate information," unless (A) the inaccurate information was included on the application "with knowledge that it was inaccurate," and (B) the inaccuracy, if known to the Register, would have caused refusal of registration. Where both conditions are alleged, § 411(b)(2) requires the court to ask the Register whether the inaccuracy would have caused refusal.

Defendants spent a decade building an invalidation practice on this provision, arguing that any material misstatement on an application — a wrong publication date, a mischaracterized single-unit publication, an overbroad claim of authorship — voided the registration and with it the plaintiff's ability to sue.

Unicolors, Inc. v. H&M Hennes & Mauritz, L.P. (2022) largely shut that practice down. Unicolors had registered thirty-one fabric designs in a single application as a "single unit of publication," a designation the Ninth Circuit held unavailable because the designs had not all been published together. The Ninth Circuit held the registration invalid, reasoning that Unicolors' lack of knowledge of the legal requirement was no excuse because § 411(b)(1)(A) excused only factual mistakes.

The Supreme Court reversed. Section 411(b)(1)(A) requires actual knowledge of inaccuracy, and "knowledge" in that provision covers both factual and legal knowledge: nothing in the statutory language suggests that a mistake of law is treated differently from a mistake of fact. A good-faith mistake about what the law requires — including the single-unit-of-publication rule — does not invalidate a registration. The Court noted the familiar maxim that ignorance of the law is no excuse but explained that it has little force where a statute conditions liability on actual knowledge, and observed that the copyright registration system is used heavily by "authors, filmmakers, artists, and composers" who "are not copyright lawyers."

Practical effect:

• For plaintiffs. An inadvertent misstatement is no longer fatal. The defense must now plead and prove that the applicant actually knew the statement was inaccurate, which is a subjective showing.
• For defendants. The viable version of the defense is a willfulness showing — evidence the applicant knew the truth and filed anyway. Circumstantial evidence of knowledge is admissible, and the Court expressly said so, so the defense survives where the record shows the applicant was told the correct rule and disregarded it. Where the claim is genuinely one of ignorance, the defense fails and the § 411(b)(2) referral to the Register never happens, because the first prong is unsatisfied.
• Pleading discipline. Because both prongs are required, a motion that establishes only materiality — that the inaccuracy would have caused refusal — does not get the defendant anywhere. Many pre-2022 briefs, and some district court opinions, treat materiality as sufficient; they are no longer good law on that point.

Limitations and accrual under § 507(b)

17 U.S.C. § 507(b) provides that a civil action must be "commenced within three years after the claim accrued." Nine words, and two Supreme Court decisions in the last decade that still leave the central question open.

Petrella: separate accrual, and no laches

Petrella v. Metro-Goldwyn-Mayer, Inc. (2014) concerned the screenplay behind Raging Bull. The author's heir gave notice of her claim in 1998, sued in 2009, and MGM argued laches barred the entire action because of the eleven-year delay. The Ninth Circuit agreed.

The Supreme Court reversed, 6–3, and established three propositions that still govern:

1. Laches cannot bar a claim for damages brought within the § 507(b) window. Congress set the time limit; a judge-made doctrine cannot shorten it. "Laches is a defense developed by courts of equity" and its application to bar relief "within a federal limitations period" would clash with the legislative judgment. The Court left a narrow door open in footnote 15 for "extraordinary circumstances" bearing on the scope of equitable relief — a permanent injunction, for instance — but not for damages.
2. The separate-accrual rule. "Each infringing act starts a new limitations period." Where a defendant commits a series of discrete infringing acts, the plaintiff may sue on each act within three years of that act, regardless of how long the course of conduct has been running. A 1980 infringement is time-barred; the identical 2023 infringement is not.
3. The three-year lookback on relief. A plaintiff who sues in year twenty can reach only the infringements occurring in the three years before suit, and can recover only the damages attributable to those acts. The Court described the rule as confining the plaintiff to "retrospective relief running only three years back from the date the complaint was filed."

Petrella also rejected the argument that a plaintiff who sits on a claim while a defendant invests in exploitation of the work should lose it. The answer was that § 507(b)'s lookback already limits the plaintiff's recovery, and that a copyright owner has no duty to sue at the first hint of infringement — she may reasonably wait to see whether the infringement is worth the expense of litigation.

Warner Chappell: no separate damages cap — and the question it left open

Petrella assumed, because the parties did not dispute it, that the claim accrued when the infringing act occurred. It did not address the discovery rule — the approach adopted by every circuit to consider it, under which a claim accrues when the plaintiff discovers, or with reasonable diligence should have discovered, the infringement. Footnote 4 of Petrella pointedly noted that the Court had "not passed on the question."

Warner Chappell Music, Inc. v. Nealy (2024) answered a narrower question and left that one open. Nealy sued in 2018 over infringements reaching back to 1983, relying on the Eleventh Circuit's discovery rule: he had been incarcerated and asserted he did not learn of the infringements until 2016. The Eleventh Circuit held the claims timely and permitted recovery for the full period. The Second Circuit had held, in Sohm v. Scholastic, that even a timely discovery-rule claim was subject to a three-year damages lookback drawn from Petrella.

The Supreme Court affirmed the Eleventh Circuit, 6–3. Be precise about the holding:

• What it decided. Where a copyright claim is timely under the discovery rule, the Copyright Act contains no separate three-year limit on damages. The Act "does not establish such a time limit"; § 507(b) addresses when a suit may be brought, and §§ 504 and 505 impose no temporal cap on recovery. Petrella's three-year-lookback language described the consequence of the facts in that case — where accrual was conceded to be occurrence-based, a plaintiff necessarily could reach only three years back — and was not a freestanding remedial rule. Sohm is therefore wrong.
• What it expressly did not decide. Whether the discovery rule governs accrual under § 507(b) at all. The Court said it was resolving the damages question "on the assumption" that the discovery rule applies, because Warner Chappell had not sought review of that premise. It dismissed as improvidently granted, in effect, the larger question. Justice Gorsuch's dissent for three Justices argued the Court should not have answered the damages question while the antecedent accrual question was unresolved, and stated the dissenters' view that the Act "almost certainly does not tolerate a discovery rule."
• What that means for practice. The live issue in every stale-infringement case is now the antecedent one: does the discovery rule apply? Three Justices have said it probably does not; every court of appeals to address it has said it does. A plaintiff relying on late discovery should expect the premise to be attacked directly, and should brief the traditional grounds for the discovery rule in federal limitations statutes — the presumption for fraud and concealment claims, the "self-concealing" character of much digital infringement, and the near-unanimous circuit authority — rather than treating Warner Chappell as having settled it.

For defendants, the sequence is the whole strategy. Attack accrual first: argue for occurrence-based accrual and, failing that, argue that a diligent plaintiff would have discovered the infringement earlier. Reasonable diligence is a fact question well suited to Rule 56 where the infringement was public and indexed — a distributed product, a searchable database, an app store listing. Once the claim is timely under the discovery rule, Warner Chappell means the damages exposure is no longer capped at three years, and the case can be worth an order of magnitude more.

Remedies and fee-shifting

17 U.S.C. § 504 gives the plaintiff an election. Under § 504(b), actual damages plus the infringer's profits attributable to the infringement and not taken into account in computing actual damages. Under § 504(c), statutory damages — provided § 412 does not bar them — of between $750 and $30,000 per work, raised to as much as $150,000 per work for willful infringement and reducible to as little as $200 where the infringer sustains the burden of proving innocent infringement. The election may be made at any time before final judgment, which in practice means a plaintiff can try the case both ways and choose after the verdict on liability.

Two recurring fights under § 504(c). The first is the number of works: statutory damages are awarded per work infringed, and "all the parts of a compilation or derivative work constitute one work," so a defendant who copies a 200-photograph database may face one award or two hundred depending on how the registration and the publication history are characterized. The second is willfulness, which is a jury question where a jury is demanded, and on which evidence of the defendant's knowledge of the registration, of prior cease-and-desist correspondence, and of any internal legal advice is decisive.

17 U.S.C. § 505 provides that the court "may" allow recovery of full costs by or against any party, and "may" award a reasonable attorney's fee to the prevailing party. Three Supreme Court decisions define the discretion.

Fogerty: no dual standard

Fogerty v. Fantasy, Inc. (1994) rejected the "dual standard" that several circuits had applied, under which a prevailing plaintiff recovered fees as a matter of course while a prevailing defendant recovered only on a showing that the suit was frivolous or brought in bad faith. John Fogerty, sued for infringing a song he had himself written, won at trial and was denied fees under that standard.

A unanimous Court held that § 505's text draws no distinction between plaintiffs and defendants, and that prevailing plaintiffs and prevailing defendants must be treated alike. The Court also refused to adopt the British Rule of automatic fee-shifting: fees under § 505 are discretionary, not automatic, and "attorney's fees are to be awarded to prevailing parties only as a matter of the court's discretion." The policy justification is the one to quote in a defense fee motion: because copyright's purpose is to enrich the public domain as well as to reward authors, "defendants who seek to advance a variety of meritorious copyright defenses should be encouraged to litigate them to the same extent that plaintiffs are encouraged to litigate meritorious claims of infringement."

Footnote 19 of Fogerty listed the nonexclusive factors that the circuits have used ever since: frivolousness, motivation, objective unreasonableness in both the factual and legal components of the case, and the need in particular circumstances to advance considerations of compensation and deterrence.

Kirtsaeng: objective reasonableness weighs heavily but does not control

Kirtsaeng v. John Wiley & Sons, Inc. (2016) — the fee decision that followed the same party's merits victory on the first-sale doctrine in Kirtsaeng v. John Wiley & Sons, Inc. (2013) — addressed how much weight the Fogerty factors carry. The Second Circuit had treated objective reasonableness as the "controlling" consideration, effectively denying fees whenever the losing side's position had been reasonable.

The Court vacated, unanimously. Objective reasonableness is "an important factor" entitled to substantial weight, because fee-shifting against a party with a reasonable position would chill exactly the litigation the Act needs to clarify its boundaries. But it "cannot be the controlling one." A district court must give due consideration to all the circumstances, and the Court identified two situations where fees may be appropriate despite a reasonable losing position: litigation misconduct, and a pattern of overaggressive assertions of copyright claims where an award serves deterrence. The remand instruction is the operative language for a fee brief: the district court must exercise discretion "in light of the considerations we have identified," giving substantial but not dispositive weight to reasonableness.

Rimini Street: "full costs" means the enumerated costs

Rimini Street, Inc. v. Oracle USA, Inc. (2019) resolved what "full costs" means in § 505. Oracle had recovered roughly $12.8 million in litigation expenses beyond taxable costs — expert witness fees, e-discovery vendor charges, jury consultant fees — on the theory that the adjective "full" expanded § 505 beyond the six categories of costs enumerated in 28 U.S.C. § 1920 and made taxable by 28 U.S.C. § 1821. The Ninth Circuit affirmed.

A unanimous Court reversed and cut the award. Sections 1920 and 1821 "define what the term 'costs' encompasses in the subject-specific federal cost statutes," and a subject-specific statute departs from that baseline only with an explicit statutory instruction. The word "full" is "a term of quantity or amount," not a term that expands the categories: it "does not alter the meaning of the word 'costs.'" A "full" award of costs is the whole of the six enumerated categories and nothing more.

The practical consequence is a large and often overlooked gap in copyright recovery. Expert fees, e-discovery hosting and processing, and consultant costs are recoverable only to the extent they can be folded into a § 505 attorney's fee award as reasonable litigation expenses billed to the client — which many district courts permit for out-of-pocket disbursements customarily billed, and which is therefore where the argument now has to be made. Budget accordingly, and do not assume a fee-and-cost award will make a prevailing party whole.

Preemption under § 301

17 U.S.C. § 301 is an express preemption provision, and an unusually broad one. As of January 1, 1978, all rights "equivalent to any of the exclusive rights within the general scope of copyright as specified by section 106" in works of authorship "fixed in a tangible medium of expression and come within the subject matter of copyright as specified by sections 102 and 103" are governed exclusively by the Act, and "no person is entitled to any such right or equivalent right in any such work under the common law or statutes of any State."

Every circuit applies a two-part test:

1. Subject matter. The work falls within the subject matter of copyright as defined by 17 U.S.C. § 102 and § 103 and is fixed in a tangible medium. Note the trap: the inquiry asks whether the work is within the subject matter, not whether it is protectable. Unoriginal compilations of facts are within the subject matter of copyright and therefore inside the preemptive field even though Feist denies them protection — which is why state-law misappropriation claims over uncopyrightable data are preempted rather than saved.
2. Equivalent rights. The state-law right asserted is equivalent to one of the § 106 exclusive rights — reproduction, preparation of derivative works, distribution, public performance, public display, digital audio transmission of sound recordings. The test the circuits use is the "extra element" test: if the state claim requires proof of an element qualitatively different from unauthorized copying, distribution, performance, or display, it is not equivalent and survives.

Applied, the sorting is fairly predictable:

The last row is the historical residue, and it explains Goldstein v. California (1973). California prosecuted record pirates under a state statute at a time when the federal Copyright Act did not protect sound recordings at all. The Court upheld the state law 5–4, holding that the Copyright Clause's grant to Congress is not an exclusive grant that by its own force displaces state power, and that Congress's decision to leave a category of writings unprotected is not necessarily a decision that the states must leave it unprotected either. Goldstein describes a pre-1976 world of concurrent federal and state authority over unpublished and federally unprotected works — the world that § 301 was enacted to end.

The one durable survivor of that world is pre-1972 sound recordings, which Congress excluded from § 301 preemption when it brought sound recordings into federal protection in 1972 and which remained subject to state law for decades. The Music Modernization Act of 2018 federalized them prospectively, creating a federal cause of action for unauthorized use of pre-1972 recordings and a schedule under which state-law protection expires. A dispute over a pre-1972 recording therefore requires checking the MMA's transition provisions rather than assuming either federal exclusivity or state authority.

A separate and independent preemption channel is conflict preemption under the Supremacy Clause, which applies where a state law interferes with the federal scheme even if § 301's two prongs are not met — the Sears/Compco line on state protection of unpatented and uncopyrighted designs. Where a § 301 argument fails on the extra-element prong, conflict preemption is the fallback, and it is the theory that does the work against state laws purporting to protect works that Congress deliberately left in the public domain.

Dastar and the copyright/trademark boundary

The most important boundary case in this area is not a copyright decision at all. Dastar Corp. v. Twentieth Century Fox Film Corp. (2003) is a Lanham Act case, and it is essential reading for anyone litigating either side of the copyright/trademark line.

Fox's copyright in a television series based on Eisenhower's Crusade in Europe had expired and the work had entered the public domain. Dastar copied the series, edited it, and sold it as its own product under its own name. Unable to sue for infringement of an expired copyright, Fox sued under 15 U.S.C. § 1125(a) — Lanham Act § 43(a) — for reverse passing off: falsely designating the origin of goods by selling another's work as one's own.

The Court rejected the claim, essentially unanimously. "Origin of goods" in § 43(a) refers to "the producer of the tangible goods that are offered for sale, and not to the author of any idea, concept, or communication embodied in those goods." Dastar was the producer of the physical videotapes it sold; that its content originated with Fox was not an origin-of-goods question. Reading § 43(a) otherwise, the Court said, would create "a species of mutant copyright law" — a perpetual, federal, attribution right that Congress had declined to enact outside the narrow visual-arts provision in 17 U.S.C. § 106A. The Court invoked the public-domain principle from the Sears/Compco line and from TrafFix Devices, Inc. v. Marketing Displays, Inc.: once a federal intellectual-property term expires, "the right to copy" passes to the public, and trademark law may not be used to reclaim it.

Dastar does three things a practitioner needs to keep straight:

• It kills the substitute claim. A plaintiff whose copyright has expired, who never registered, who is barred by § 412 or § 507(b), or who lost on fair use cannot re-plead the same grievance as a false-designation claim. This is the single most common use of Dastar on a motion to dismiss.
• It does not reach false advertising. Section 43(a)(1)(B) claims about the nature, characteristics, or qualities of goods survive Dastar, and POM Wonderful LLC v. Coca-Cola Co. confirms § 43(a) has independent force where another federal scheme is in play. A defendant who misrepresents what a product is may still be liable; a defendant who merely fails to credit an author generally may not.
• It leaves the ordinary trademark case untouched. Where the mark itself — a title with secondary meaning, a character name used as a source identifier, trade dress — signals commercial source rather than authorship, the Lanham Act applies normally. The overlap analysis and the Rogers framework are treated in the federal trademark jurisdiction guide.

Sovereign immunity and government works

Allen v. Cooper: no damages against a state

Blackbeard's flagship Queen Anne's Revenge sank off North Carolina in 1718. A salvage company documented the recovery and registered copyrights in the resulting video and photographs; North Carolina published the material, and when the company sued, the State pleaded sovereign immunity. Congress had anticipated exactly this: the Copyright Remedy Clarification Act of 1990 provides that a state "shall not be immune, under the Eleventh Amendment … or under any other doctrine of sovereign immunity, from suit in Federal court" for copyright infringement.

Allen v. Cooper (2020) held the CRCA invalid. Congress may abrogate state sovereign immunity only with a clear statement — satisfied here — and with a valid constitutional authorization, which was not.

• The Intellectual Property Clause does not supply it. Florida Prepaid Postsecondary Education Expense Board v. College Savings Bank had held in 1999 that the materially identical Patent Remedy Act was not supported by Article I, following Seminole Tribe's rule that Article I powers do not authorize abrogation. The Court applied stare decisis: the Copyright Clause and the Patent Clause are the same clause, and the two statutes were "basically identical." Central Virginia Community College v. Katz, the bankruptcy exception, was confined to the "singular" nature of the Bankruptcy Clause.
• Section 5 of the Fourteenth Amendment does not supply it on this record. Abrogation under § 5 requires congruence and proportionality between the remedy and a documented pattern of unconstitutional conduct — here, state infringement that amounts to an uncompensated taking of property without due process. Negligent infringement is not a constitutional violation, and a state's infringement is not a due process violation where state law provides an adequate remedy. The legislative record behind the CRCA showed only a dozen or so examples of state infringement over a period of years, with scant evidence of intentional conduct. Against that record a blanket abrogation covering every state and every infringement was "too far" out of proportion.

The Court closed with an explicit invitation: a "tailored statute" resting on a proper legislative record of intentional or reckless state infringement "could" pass § 5 muster. Congress has not enacted one, and the Copyright Office's 2021 study on the subject produced no legislation.

The practical upshot is that a state actor's copyright infringement is frequently unremediable in damages. What remains:

• Prospective injunctive relief against the responsible state officer under Ex parte Young. This is the principal live remedy and should be pleaded against individual officials in their official capacities, seeking an injunction rather than retrospective compensation.
• Damages against individual officials personally, where the doctrine of the forum permits and qualified immunity can be overcome — difficult, but not always foreclosed.
• Takings claims in state court or under state inverse-condemnation procedures, which Allen itself contemplates as the adequate state remedy that defeats the due process theory.
• Contract remedies, which sovereign immunity does not reach in the same way where the state has waived immunity for contract claims — often the strongest available theory, and a reason to get a license or a written agreement with any state entity before delivering a work.
• Note the asymmetry. Municipalities, counties, school districts, and other political subdivisions are not protected by Eleventh Amendment immunity, so infringement by a city or a local school board is fully actionable. The arm-of-the-state inquiry is therefore often the whole case. State universities generally are arms of the state; local community colleges and municipal libraries frequently are not.

Georgia v. Public.Resource.Org and the government edicts doctrine

Georgia v. Public.Resource.Org, Inc. (2020) is the other half of the state-and-copyright picture, and the case most directly about access to legal materials. Georgia's official code, the OCGA, was published with annotations — summaries of judicial decisions, editorial notes, research references — prepared by Lexis under a work-for-hire agreement with the State's Code Revision Commission, with the copyright vesting in the State. Public.Resource.Org posted the entire annotated code online. Georgia sued for infringement.

The Court held, 5–4, that the annotations are uncopyrightable under the government edicts doctrine. The rule from the nineteenth-century cases — Wheaton v. Peters, Banks v. Manchester, Callaghan v. Myers — is author-centric rather than content-centric: judges and legislators cannot be the "authors" of the works they produce in the course of their official duties, whatever those works are. Because the Code Revision Commission is an arm of the Georgia legislature and produced the annotations in the discharge of its legislative duties, the annotations have no author for copyright purposes, and it does not matter that they carry no legal force.

The majority rejected two arguments that had persuaded the dissent. First, the argument from the work-for-hire provisions: the Commission's authorship is the relevant authorship, and a state cannot acquire by contract an authorship that the doctrine denies it. Second, the argument that only materials with the force of law are edicts: the doctrine turns on who created the work, not on whether it binds. Chief Justice Roberts' framing of the stakes is the quotable passage — a contrary rule would produce "a Georgia citizen interested in whether his side yard is subject to a setback requirement" who must either "read the economy-class version of the Georgia Code" or pay for the "first-class version."

For practice, the holding draws a clean line. Material produced by legislators or judges in the course of their duties — statutes, judicial opinions, and legislatively authored annotations — is in the public domain. Material produced by non-legislative, non-judicial state actors, or by private standards bodies later incorporated by reference into law, is not automatically outside copyright; those cases turn on the separate and unresolved incorporation-by-reference question, which the Court did not reach.

Substantive doctrines that decide who may sue

Three substantive doctrines come up so reliably at the threshold of a copyright case that they belong in a jurisdiction-and-procedure guide.

Feist: originality is constitutional, and sweat of the brow is dead

Feist Publications, Inc. v. Rural Telephone Service Co. (1991) held that a white-pages telephone directory — names, towns, and numbers arranged alphabetically — contains no copyrightable expression. The holding rests on two propositions:

1. Originality is a constitutional requirement, not merely a statutory one. The Copyright Clause's words "authors" and "writings" "presuppose a degree of originality," and Congress cannot dispense with it. Originality requires independent creation plus "at least some minimal degree of creativity" — a low bar, but not a nonexistent one, and alphabetical arrangement of facts does not clear it.
2. Facts are never original, because the first person to find a fact has not created it. "The discoverer merely finds and records." A compilation of facts may be protected in its selection, coordination, and arrangement under § 101 and § 103, but the protection is thin and extends only to the original contribution, not to the underlying data.

What makes Feist worth flagging in a practice guide is that it rejected a doctrine the lower courts had applied for decades. "Sweat of the brow" — the theory that industrious collection of information earns copyright protection, traceable to the 1922 Eighth Circuit decision in Jeweler's Circular Publishing Co. v. Keystone Publishing Co. — had been the governing rule in most circuits. The Court called it "a theory [that] flouted basic copyright principles" and explained that it had "eschewed the most fundamental axiom of copyright law — that no one may copyright facts or ideas." A unanimous Court erased a body of authority that had been briefed and applied as settled law since the 1920s.

Modern significance: Feist is the reason database, listing, and data-scraping claims fail, and why the plaintiffs in that space plead contract, trade secret, and CFAA theories instead — where, as noted above, § 301 preemption often does not reach them.

CCNV v. Reid: work for hire and the agency test

Who owns the copyright determines who may sue, and 17 U.S.C. § 201 vests ownership initially in the author — except that for a work made for hire, the employer or commissioning party "is considered the author." The § 101 definition has two branches: (1) a work prepared by an employee within the scope of employment; or (2) one of nine enumerated categories of specially ordered or commissioned works, and only if there is a written instrument signed by both parties expressly designating the work as a work for hire.

Community for Creative Non-Violence v. Reid (1989) resolved a four-way circuit split over the meaning of "employee" in the first branch. CCNV commissioned a sculptor to create a nativity-style sculpture of homeless figures for a Washington display; when the parties fell out over ownership, CCNV argued the work was made for hire because it had supervised and directed the sculptor's work.

A unanimous Court held that "employee" carries its common-law agency meaning. The Court rejected the three alternatives that circuits had adopted — actual control over the work, right to control, and formal salaried employment — and adopted the Restatement (Second) of Agency § 220 multifactor analysis:

• the hiring party's right to control the manner and means of creation;
• the skill required;
• the source of the instrumentalities and tools;
• the location of the work;
• the duration of the relationship;
• whether the hiring party may assign additional projects;
• the extent of the hired party's discretion over when and how long to work;
• the method of payment;
• the hired party's role in hiring and paying assistants;
• whether the work is part of the hiring party's regular business;
• whether the hiring party is in business at all;
• the provision of employee benefits; and
• the tax treatment of the hired party.

The Court emphasized that no single factor is determinative, but the post-Reid case law has given outsized weight to the last three — benefits, tax treatment, and payroll status — because they are objective and documentary. Reid was an independent contractor: highly skilled, supplying his own tools, working in his own studio for two months, paid a lump sum with no benefits and no withholding. Because the sculpture also did not fall within any of the nine enumerated categories, CCNV could not claim work-for-hire authorship; the Court remanded on whether it was a joint work.

The practical lesson for litigators is that ownership is often the case. A plaintiff suing on a work created by a freelancer, a consultant, or a contractor without a signed written assignment may not own the copyright at all, and the § 101(2) written-instrument requirement is strictly enforced. A defendant's first document request should be the chain of title.

Star Athletica: separability for useful articles

Star Athletica, LLC v. Varsity Brands, Inc. (2017) resolved a long and messy circuit disagreement over when a design feature of a useful article is protectable. Section 101 protects the "pictorial, graphic, or sculptural features" of a useful article only if they "can be identified separately from, and are capable of existing independently of, the utilitarian aspects of the article." Varsity Brands claimed copyright in arrangements of chevrons, stripes, and color blocks on cheerleading uniforms.

The Court held, 6–2, that the test is purely statutory and has two steps: a feature incorporated into a useful article is eligible for copyright if it (1) can be perceived as a two- or three-dimensional work of art separate from the useful article, and (2) would qualify as a protectable pictorial, graphic, or sculptural work — either on its own or fixed in some other tangible medium — if imagined separately from the useful article.

The Court expressly discarded the accumulated apparatus: the physical/conceptual separability distinction, the objective-necessity test, the design-process inquiry into whether aesthetic choices were influenced by function, and the marketability test. It also rejected the argument that a feature is unprotectable because, once imagined separately, it would replicate the shape of the useful article — the relevant question is whether the feature is art, not whether the imagined result looks like a uniform. Applied, the surface decorations on the uniforms satisfied both steps, because they could be imagined as a two-dimensional design applied to a canvas.

The effect was to expand the field for fashion, furniture, product design, and industrial design claims substantially, because the two-step test is easier to satisfy than most of the tests it replaced. The countervailing limits are that the copyright reaches only the surface design and not the article's shape or cut, and that Feist originality still applies — simple stripes and chevrons must clear the minimal-creativity bar, which is where the defense now concentrates.

Fair use

17 U.S.C. § 107 codifies fair use as a four-factor inquiry: (1) the purpose and character of the use, including whether it is of a commercial nature or for nonprofit educational purposes; (2) the nature of the copyrighted work; (3) the amount and substantiality of the portion used in relation to the whole; and (4) the effect of the use upon the potential market for or value of the work. The statute directs that all four be weighed together, and the Court has repeatedly said the list is illustrative rather than exhaustive.

Procedurally, fair use is an affirmative defense on which the defendant bears the burden. It is a mixed question — the underlying facts are for the jury where a jury is demanded, the ultimate determination is legal — which means a fair-use ruling is available on Rule 56 where the facts about the use are undisputed, and is reviewed de novo on appeal. That reviewability is a significant strategic feature of the defense: an adverse jury finding on fair use is far more vulnerable on appeal than an adverse finding on copying.

Sony: time-shifting and substantial noninfringing use

Sony Corp. of America v. Universal City Studios, Inc. (1984) — the Betamax case — held 5–4 that private, noncommercial time-shifting of free broadcast television is fair use, and that the manufacturer of a device capable of substantial noninfringing uses is not contributorily liable merely because the device can also be used to infringe. The staple-article-of-commerce rule was borrowed from patent law, and it sheltered the consumer-electronics and computing industries for two decades.

Two pieces of Sony still do work. First, the presumption language: a "commercial" use is presumptively unfair and a noncommercial use requires the plaintiff to show market harm, a framework the Court later narrowed. Second, the market-harm formulation — the plaintiff must show "by a preponderance of the evidence that some meaningful likelihood of future harm exists" — which remains the plaintiff's burden on the fourth factor when the use is noncommercial.

Harper & Row: unpublished works and the fourth factor

Harper & Row, Publishers, Inc. v. Nation Enterprises (1985) concerned a magazine's publication of roughly 300 words of verbatim quotation from President Ford's unpublished memoir, obtained from a purloined manuscript, timed to scoop an authorized Time excerpt. Time cancelled its contract.

The Court held the use unfair, 6–3, and made two lasting contributions. First, the unpublished nature of a work is "a critical element of its 'nature'" and weighs heavily against fair use, because the author's right of first publication implicates control over when and whether to publish at all. Congress later softened this in a 1992 amendment adding the final sentence of § 107 — that a work's unpublished status "shall not itself bar a finding of fair use" if the factors are otherwise satisfied — but the factor still weighs against the defendant. Second, the Court called the fourth factor "undoubtedly the single most important element of fair use," a phrase that dominated the analysis for a decade and that Campbell later disclaimed as a mechanical hierarchy.

Harper & Row is also the source of the rule that the amount taken is measured qualitatively as well as quantitatively: 300 words out of 200,000 was fatal because they were "the heart of the book."

Campbell: parody, transformativeness, and no commercial presumption

Campbell v. Acuff-Rose Music, Inc. (1994) is the case that reorganized modern fair use. 2 Live Crew's rap version of Roy Orbison's "Oh, Pretty Woman" was held potentially fair, and the Court reversed the Sixth Circuit's contrary ruling unanimously.

Four holdings:

1. Transformativeness became the central first-factor question. Borrowing from Judge Leval, the Court 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." The more transformative the use, the less significance other factors — including commercialism — carry.
2. No commercial-use presumption. Sony's language had hardened in the lower courts into a near-dispositive presumption against commercial uses. Campbell rejected it: commerciality is one element to weigh, not a presumption, and treating it as dispositive "would swallow nearly all of the illustrative uses listed in the preamble" of § 107, since news reporting, criticism, and comment are ordinarily done for profit.
3. No presumption of market harm from a commercial transformative use; and the fourth factor considers only harm to the market for the original and its derivatives, not the "biting criticism" that suppresses demand. A parody that kills the market by ridicule does no cognizable harm.
4. Parody versus satire. Parody targets the original and therefore needs to borrow enough to conjure it up; satire uses the original to comment on something else and stands on its own feet, so it must "justify the very act of borrowing." This distinction is still the fault line in most parody cases.

Campbell also corrected Harper & Row's "single most important element" language, insisting the factors be weighed together "in light of the purposes of copyright" rather than ranked.

Google v. Oracle: the API declarations case

Google LLC v. Oracle America, Inc. (2021) held 6–2 that Google's copying of roughly 11,500 lines of Java SE API declaring code into Android was fair use. The Court assumed without deciding that the declarations were copyrightable, which is the first thing to note — the copyrightability of API structure remains formally unresolved.

The analysis is idiosyncratic but the reasoning matters:

• Nature of the work. The declaring code is "inherently bound together with uncopyrightable ideas" and "further than most computer programs" from the core of copyright, because its value lies in the investment of third-party programmers who learned it rather than in its own expressive content.
• Purpose and character. Reimplementing the declarations to allow programmers to carry over their existing skills into a new, transformative computing platform — smartphones — was a transformative purpose, even though Android is enormously commercial.
• Amount. The 11,500 lines were measured against the 2.86 million lines of the whole API, and the Court stressed that Google took only what was needed to allow programmers to work in a familiar language.
• Market effects. The Court credited jury findings that Java SE's smartphone market was not one Oracle was positioned to serve, treated Oracle's lost licensing revenue as partly attributable to the programmers' own investment, and — importantly for later cases — instructed that courts consider the public benefits of the copying, including the risk that enforcement would let the copyright holder "limit the future creativity of new programs."

The Court also addressed procedure: the Federal Circuit had reversed a jury verdict of fair use, and although the Court held that fair use is a legal question reviewed de novo, it emphasized that subsidiary factual findings are for the jury and must be respected.

Warhol: the retrenchment

Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith (2023) is the most consequential fair-use decision in thirty years, and its importance lies in what it did to the reading of Campbell that the lower courts had adopted.

Lynn Goldsmith photographed Prince in 1981. Warhol used the photograph, under a license limited to a single Vanity Fair illustration, to create a series of silkscreens. In 2016, after Prince's death, the Warhol Foundation licensed Orange Prince to Condé Nast for a magazine cover — the same use, in the same market, that Goldsmith licenses her photographs for.

The Court held 7–2 that the first factor favored Goldsmith. The holding is narrow on its face and broad in effect:

• The inquiry is use-specific. The first factor asks about "the specific use of a copyrighted work that is alleged to be an infringement," not about the new work in the abstract. The question was not whether Orange Prince is a distinct artwork — it plainly is — but whether the Foundation's 2016 commercial licensing of it to a magazine served a purpose sufficiently distinct from Goldsmith's licensing of the same photograph to a magazine. It did not: both were portraits of Prince used to illustrate stories about Prince.
• Transformativeness is a matter of degree and is not satisfied by added meaning alone. "Adding new expression, meaning, or message" is necessary but not sufficient. If mere new meaning sufficed, the derivative-work right in § 106(2) — which by definition covers works that recast, transform, or adapt the original — would be swallowed whole. The Court insisted that the transformativeness inquiry be conducted with an eye to that structural constraint.
• Commercial substitution is the pivot. The first factor is "relative to" the degree of commercial character, and the driving concern is whether the challenged use substitutes for the original in a market the original serves. Where the two uses "share substantially the same purpose" and the use is commercial, the first factor disfavors fair use "without a compelling justification."
• Justification must be supplied. Copying that targets the original — parody, criticism, commentary about the original itself — carries its own justification. Copying that merely uses the original as raw material for an unrelated message does not, and must be justified.

The practical retrenchment from the broad reading of Campbell is substantial. For two decades the Second Circuit in particular — Cariou v. Prince, and the Second Circuit's own decision in this case before it was reversed — had asked whether a new work reasonably could be perceived as conveying a different aesthetic or message, an inquiry that a competent art historian could satisfy for almost any appropriation. Warhol replaced that with a market-substitution question anchored to the specific challenged use. Justice Gorsuch's concurrence made the point plainly: the statute asks about the use, not about the artist's intent or the work's meaning. Justice Kagan's long dissent for two Justices accused the majority of misreading Campbell and of leaving "no room" for the transformative use of existing images.

For briefing, the sequence after Warhol is: identify the specific use challenged, characterize the plaintiff's own licensing market with evidence, and argue substitution or non-substitution within that market. A first-factor argument that consists of asserting new meaning, without market analysis, no longer wins.

Secondary liability and technology

The Copyright Act says nothing about contributory or vicarious liability; both are judge-made, and both are imported from tort and agency principles. Contributory liability requires knowledge of the infringing activity plus a material contribution to it. Vicarious liability requires the right and ability to supervise plus a direct financial interest in the infringement, and requires no knowledge at all.

Grokster: inducement

Metro-Goldwyn-Mayer Studios Inc. v. Grokster, Ltd. (2005) addressed peer-to-peer file-sharing software whose overwhelming actual use was infringement. The Ninth Circuit had read Sony to immunize any technology capable of substantial noninfringing use, without regard to the distributor's conduct.

A unanimous Court reversed and recognized a distinct inducement theory: "one who distributes a device with the object of promoting its use to infringe copyright, as shown by clear expression or other affirmative steps taken to foster infringement, is liable for the resulting acts of infringement by third parties." Sony was not overruled or even reinterpreted; the Court held only that Sony's staple-article rule "was never meant to foreclose rules of fault-based liability derived from the common law," and that where the record shows culpable intent, capability of lawful use is no defense.

The evidence the Court relied on is the template for every inducement case since: internal communications aiming at the former Napster userbase, advertising and branding directed at infringing use, the absence of any filtering or other effort to diminish infringement, and a business model whose advertising revenue scaled directly with infringing volume. Note the last two carefully — the Court said failure to filter and a volume-based revenue model are not independently sufficient, but are probative when combined with expressions of unlawful purpose.

Aereo: the public performance right, decided by analogy

American Broadcasting Cos. v. Aereo, Inc. (2014) is the clearest recent example of the Court resolving a technology dispute by analogy rather than by first principles. Aereo assigned each subscriber an individual dime-sized antenna, made an individual personal copy of the broadcast signal, and streamed that individual copy to that individual subscriber — an architecture designed around the Second Circuit's Cablevision decision to avoid the public performance right in § 106(4) and the transmit clause in § 101.

The Court held 6–3 that Aereo publicly performed. The reasoning was historical and functional. The 1976 Act's transmit clause was enacted specifically to override Fortnightly and Teleprompter, two decisions holding that community antenna systems did not "perform" broadcasts. Because Aereo's service is "substantially similar to those of the CATV companies that Congress amended the Act to reach" — because it "looks like cable television" — the technical differences behind the user's screen do not matter. The subscriber-specific copies were "invisible" to the user and were an artifact of a design choice made for legal rather than functional reasons.

The Court also tried to cabin the holding: it emphasized that its reasoning did not necessarily apply to cloud storage or remote-storage services, that an entity that merely supplies equipment may not perform, and that it was not addressing services in which the user supplies the content. Justice Scalia's dissent for three Justices argued that the majority had adopted an "improvised standard" — "looks-like-cable-TV" — with no statutory basis, and that volitional conduct is the correct framework: Aereo's system was user-directed, so Aereo did not itself perform.

What the analogical method means for the AI cases

Sony, Grokster, and Aereo together describe how the Court handles technologies Congress did not anticipate: it identifies the closest analogue in the existing scheme and asks whether the new service occupies the same functional position. It does not build a doctrine from the statute's text alone, and it has been willing to look past architecture chosen for legal reasons.

That method is the backdrop for the generative-AI training litigation now pending across the district courts — the authors' and publishers' suits against model developers, the visual-artist suits, the news-organization suits, and the music-publisher actions. The recurring questions are whether ingesting copyrighted works to train a model is a use whose purpose is distinct from the original within the meaning of Warhol, whether the relevant market harm is lost licensing revenue for training data or substitution in the market for the outputs, whether intermediate copying is excused on the reasoning of the reverse-engineering line and Google v. Oracle, and whether a model's parameters are themselves a copy or a derivative work. District courts have so far divided, with some decisions treating training as transformative and others emphasizing the emerging licensing market for training corpora as cognizable fourth-factor harm, and several turning on whether the training copies were acquired lawfully.

Three practical points follow for anyone briefing in that space. First, Warhol is the more important citation than Campbell: the question a court will ask is what the specific use is and whether it substitutes in a market the plaintiff serves. Second, Google v. Oracle's public-benefit reasoning and its treatment of intermediate copying are the strongest defense authority, and its narrow framing — it assumed copyrightability and stressed the peculiar nature of declaring code — is the strongest plaintiff response. Third, the registration and accrual sections of this guide are doing quiet work in these cases: class definitions turn on § 411(a) registration status, and the reach-back exposure turns entirely on the discovery-rule question Warner Chappell left open.

Standing, joinder, and supplemental jurisdiction

Who may sue under § 501(b)

17 U.S.C. § 501(b) confers the right to sue on "the legal or beneficial owner of an exclusive right under a copyright," and only "with respect to the particular right owned." This is a statutory standing requirement layered on top of Article III, and it disposes of a surprising number of cases.

• Legal owner — the author, or an assignee of an exclusive right by a writing satisfying § 204(a). Section 204(a) requires a signed written instrument for any transfer of copyright ownership; an oral assignment transfers nothing, though a later writing may confirm an earlier oral agreement between the parties themselves.
• Beneficial owner — the canonical example is an author who assigned the copyright in exchange for royalties, who retains standing to sue because the infringement reduces the royalty stream.
• Exclusive licensee — has standing, because an exclusive license is a transfer of copyright ownership under the § 101 definition, and the licensee owns the exclusive right within the scope of the license. The standing is limited to that scope: an exclusive licensee for hardcover publication in North America cannot sue over an unauthorized film.
• Nonexclusive licensee — has no standing, ever. A nonexclusive license is not a transfer of ownership, does not require a writing, and conveys no right that anyone can infringe. A nonexclusive licensee whose market is being destroyed by a third-party infringer has no copyright claim; its recourse is to persuade the owner to sue, or to have negotiated a right to compel enforcement in the license.
• Assignee of accrued claims — a bare assignment of the right to sue for past infringement, unaccompanied by any transfer of an exclusive right, is generally insufficient. Several circuits require that the assignment convey an exclusive right; others permit an assignment of accrued claims together with a concurrent transfer. This is a live circuit disagreement and is worth checking before acquiring claims for enforcement.

Co-owners and joinder

Joint authors are tenants in common. Each co-owner may independently license the work non-exclusively, may independently sue for infringement, and owes the others a duty to account for profits. The consequences follow from that:

• A co-owner is not an indispensable party under Rule 19 in most circuits, because one co-owner may sue alone. The better practice is nonetheless to join co-owners or obtain their consent, since a defendant will otherwise argue prejudice and a nonjoined co-owner may license the defendant mid-case.
• A co-owner cannot infringe the jointly owned copyright — the claim between co-owners is an accounting, which is a state-law claim and does not arise under the Act. This is a common jurisdictional trap: a co-author who sues a co-author for "infringement" is pleading a claim that belongs in state court unless authorship itself is disputed.
• A defendant holding a license from any one co-owner is not an infringer. Establishing a license from a single co-owner is frequently the cheapest complete defense available, and is a reason to investigate the full authorship history before answering.

Supplemental jurisdiction over the related state claims

Copyright complaints routinely carry state-law counts — breach of the same license, breach of fiduciary duty, unfair competition, unjust enrichment, right of publicity. 28 U.S.C. § 1367 supplies jurisdiction over any claim so related to the copyright claim as to form part of the same case or controversy under Article III, which the common-nucleus standard of United Mine Workers of America v. Gibbs governs. Because § 1338(a) exclusivity means the federal claim can only be heard federally, the § 1367 analysis is straightforward in the ordinary case.

Two cautions. First, § 301 preemption will kill many of the appended state claims on the merits, so pleading them creates a target rather than an alternative. Screen each one against the extra-element test before including it. Second, § 1367(c) discretion is real: where the copyright claim is dismissed early — for want of registration under § 411(a), for instance — the state claims will usually go with it, and § 1367(d) will have suspended the state limitations clock in the meantime under Artis v. District of Columbia. Section 1338(b) separately grants jurisdiction over a related unfair-competition claim joined with a substantial copyright claim, which predates § 1367 and remains available.

The § 411(a) notice to the Register

Two distinct notice obligations run to the Register of Copyrights and are frequently confused.

1. Suit after refusal. Where the Register has refused registration, § 411(a) permits the action to proceed only if "notice thereof, with a copy of the complaint, is served on the Register of Copyrights." The Register may then intervene on the registrability question within sixty days; failure to intervene does not deprive the court of jurisdiction to determine the issue. Serving this notice is a mechanical step that a plaintiff proceeding after refusal must not omit.
2. Constitutional challenges. Where a party puts the constitutionality of an Act of Congress at issue — the CRCA in Allen v. Cooper, the term extension in the Eldred line, the restoration provisions in Golan — Rule 5.1 requires the party to file a notice of constitutional question and serve the Attorney General, and requires the court to certify under 28 U.S.C. § 2403. This is separate from the § 411(a) notice, runs to the Attorney General rather than to the Register, and is independently enforceable — a court may not hold a statute unconstitutional before certification.

The § 411(b)(2) referral is a third, different mechanism: where a party alleges a knowing and material registration inaccuracy, the court shall request the Register to advise whether the inaccuracy would have caused refusal. After Unicolors, the referral is triggered only if the knowledge prong is satisfied, so a court may — and regularly does — decline to refer where the record shows only an innocent mistake.

Practice checklist

1. Register before you think about suing, and file the application even earlier. Fourth Estate means no complaint until the Register acts. Section 410(d) means the effective date is the day the Office received a complete application, so early filing locks in the § 412 date even while the certificate is pending. For a client with continuing output, put routine registration on a calendar — quarterly at worst.
2. Run the § 412 calculation before you evaluate the case. Identify the date the infringement commenced, the date of first publication, and the effective registration date. If infringement commenced before registration and outside the three-month grace period, there are no statutory damages and no fees, and the case is an actual-damages case with a self-funded budget.
3. If litigation is imminent and the application is pending, file for special handling. Pay the fee and document the litigation basis. Weeks of Copyright Office processing are cheaper than a dismissal and a refiling.
4. Characterize the claim deliberately. Before filing, decide whether you are pleading infringement or breach, and plead consistently. An infringement complaint needs unauthorized exercise of a § 106 right; a license-scope theory needs the breached clause to be a condition of the grant, not an independent covenant. A complaint that reads like a contract dispute invites a § 1338 dismissal after discovery.
5. Prove the chain of title in the first pleading. Section 501(b) standing requires legal or beneficial ownership of the specific exclusive right at issue. Check for a § 204(a) writing on every transfer, confirm whether contractor-created works were validly assigned rather than assumed to be works for hire under Reid, and identify every co-owner. As a defendant, request the chain of title immediately and look for a license from any single co-owner.
6. Brief accrual before you brief anything else on limitations. Petrella gives separate accrual and kills laches; Warner Chappell removes the three-year damages cap for claims timely under the discovery rule but expressly reserves whether the discovery rule exists. Plaintiffs should support the discovery rule affirmatively; defendants should attack it and, in the alternative, attack reasonable diligence on a public, indexed infringement.
7. Screen every appended state-law count against § 301. Apply the two-part test and the extra-element rule. Keep contract, trade secret, publicity, and fraud theories; drop misappropriation, conversion of intangibles, and unjust enrichment premised on copying, which will be preempted and will cost you a motion.
8. *Use Dastar on any Lanham Act count that is really a copyright claim. A false-designation-of-origin theory premised on uncredited authorship fails under Dastar*. Preserve the distinction for genuine § 43(a)(1)(B) false-advertising claims about what a product is.
9. Check whether the defendant is an arm of the state before valuing the case. After Allen v. Cooper, a state defendant's damages exposure is effectively zero; plead Ex parte Young injunctive relief against officers, look for a contract waiver, and confirm whether the entity is really a state agency rather than a municipality or local district, which enjoy no immunity.
10. Frame fair use around the specific use, with market evidence. After Warhol, the first factor turns on whether the challenged use substitutes in a market the plaintiff serves. Build the record on the plaintiff's actual licensing practice — license agreements, rate cards, comparable deals — and argue substitution or non-substitution from it. Do not rest on added meaning.
11. Budget costs realistically. Rimini Street limits § 505 "full costs" to the § 1920 and § 1821 categories. Expert fees and e-discovery expenses are recoverable only if you can fold them into a reasonable attorney's-fee award, so structure the billing records and the fee application with that in mind from the start.
12. Serve the required notices. If you are suing after a refusal, serve the Register with notice and a copy of the complaint under § 411(a). If any party challenges the constitutionality of a federal statute, file the Rule 5.1 notice and obtain the § 2403 certification.

Frequently asked questions

Can a copyright case ever be litigated in state court?

Not as a copyright claim. Section 1338(a) makes federal jurisdiction over copyright claims exclusive, so a state court cannot adjudicate infringement. State courts do, however, routinely decide disputes about copyrights — contract, ownership under an assignment, royalty accounting between co-owners, employment agreements — and Gunn v. Minton confirms that a state court's resolution of an embedded federal intellectual-property question is legitimate. There is also no copyright analogue to the patent removal provision in § 1454, so a copyright counterclaim asserted in a state contract action cannot be removed.

My client's registration application has been pending for eight months and the infringement is ongoing. What can we do?

File for special handling and pay the expedite fee, citing the prospective litigation. Meanwhile, send a cease-and-desist letter — that is not a "civil action" and § 411(a) does not restrict it, and the letter also builds the willfulness record for § 504(c). Consider whether a non-copyright theory supports immediate relief: a contract claim, a trade secret claim, or a § 43(a)(1)(B) false-advertising claim if the defendant is misrepresenting the product. Do not file a complaint on the unregistered work hoping to amend later; Fourth Estate is unambiguous, and the dismissal will be with leave but will cost the schedule.

Does a defective registration end the case?

Almost never after Unicolors. Section 411(b) invalidates a registration only if the applicant included the inaccurate information with actual knowledge it was inaccurate and the inaccuracy would have caused refusal. Unicolors held that a lack of either factual or legal knowledge excuses the error, so a good-faith mistake about what the law requires — including the single-unit-of-publication rule at issue there — leaves the registration intact. A defendant needs evidence of knowing misstatement, which may be circumstantial, and a materiality showing alone gets nowhere.

Is § 411(a) jurisdictional?

No. Reed Elsevier v. Muchnick held it is a claim-processing precondition rather than a limit on subject-matter jurisdiction, placing it in the Arbaugh line. The practical consequences cut both ways: a registration objection is forfeited if not timely raised and cannot be resurrected on appeal, but it remains a mandatory precondition that entitles a defendant who does raise it to dismissal of the claims on unregistered works.

Exactly what did Warner Chappell decide?

One thing only: where a copyright claim is timely under the discovery rule, the Copyright Act imposes no separate three-year cap on damages, so a plaintiff who timely discovers an old infringement may recover for the full period of infringement. The Court reached that result on the assumption that the discovery rule governs accrual, and expressly declined to decide whether it does — a premise three dissenting Justices said is almost certainly wrong. Every court of appeals to address accrual has adopted the discovery rule, but the question is now openly contested and must be briefed on its own.

Can we still argue the plaintiff waited too long?

Not as laches, at least against damages. Petrella holds that laches cannot bar a damages claim filed inside the § 507(b) window, and that each infringing act starts its own clock under the separate-accrual rule. The surviving arguments are that the claim is untimely on its own terms — occurrence-based accrual, or failure of reasonable diligence under the discovery rule — and, under Petrella's footnote on extraordinary circumstances, that delay should shape the scope of equitable relief such as a permanent injunction or an order affecting a defendant's continuing investment.

When are fees available under § 505, and how much of the litigation cost do they cover?

Fees are discretionary for either side on equal terms. Fogerty rejected the dual standard and refused to make fees automatic; Kirtsaeng holds that objective reasonableness carries substantial weight but is not controlling, and that fees may be appropriate despite a reasonable position where there was litigation misconduct or a pattern of overaggressive claim assertion. On costs, Rimini Street limits "full costs" to the six categories in § 1920 as made taxable by § 1821, so the $12.8 million in expert and e-discovery expenses awarded there was reversed. Those expenses are recoverable, if at all, only as part of a reasonable attorney's-fee award.

How much does state sovereign immunity actually protect a state infringer?

Substantially. Allen v. Cooper held the Copyright Remedy Clarification Act invalid, so a state cannot be sued for copyright damages in federal court — neither the Intellectual Property Clause nor § 5 of the Fourteenth Amendment supported abrogation on the record Congress compiled, following the parallel patent holding in Florida Prepaid. What remains is prospective injunctive relief against state officers under Ex parte Young, possible individual-capacity claims, state takings or inverse-condemnation remedies, and contract claims where the state has waived immunity. Municipalities and local districts have no Eleventh Amendment immunity, so the arm-of-the-state question is often decisive.

What does Warhol mean for an appropriation-art or AI-training defense?

It shifts the first factor from meaning to markets. The question is no longer whether the new work can be perceived as conveying a different message, which is how many courts had read Campbell, but whether the specific challenged use shares substantially the same purpose as the original and substitutes for it commercially. A defense brief now has to characterize the plaintiff's licensing market with evidence and show non-substitution, or supply a justification for copying that targets the original itself. Google v. Oracle remains the strongest authority for intermediate and interoperability copying and for weighing public benefits, but its narrow framing — it assumed copyrightability and stressed the unusual nature of declaring code — limits how far it travels.

This guide covers the law as of September 2026. Every case and statute above links to its full text on LawDiver.