OUR SERVICES
How software copyright infringement cases over source code actually work, from the registration gate to the substantial similarity fight, whether you are enforcing or defending.
A competitor’s product suddenly does what yours does, the way yours does it. A former engineer’s new venture ships in months what took your team years. Or the letter comes to you: a vendor claims your team copied its code, and it wants an injunction. Either way, you are in a software copyright dispute. These cases follow their own playbook, one that looks very little like other intellectual property or commercial litigation.
Software copyright infringement means copying protected elements of a computer program without permission. The catch is that copyright law protects some parts of a program and not others. Nearly every software copyright case is really a fight about where that line falls. Understanding the playbook early, before positions harden and evidence scatters, is most of the battle.
A software copyright plaintiff has to prove two things. First, that it owns a valid copyright in the program. Second, that the defendant copied protected elements of it. Copying, in this sense, means exercising one of the owner’s exclusive rights under the Copyright Act without permission, such as reproducing, adapting, distributing, or displaying the work. Unauthorized copying is rarely admitted, so it is usually shown by access plus substantial similarity between the works. Where two programs are strikingly similar, courts can infer the copying without direct proof of access. In software cases access is often conceded, because the defendant licensed the product, worked on the code, or downloaded it. That concentrates the entire case on substantial similarity, which is where the technical fight lives.
There is also a gate before the courthouse door. Copyright claims belong in federal court, and with narrow exceptions you cannot file an infringement lawsuit there until the Copyright Office has actually registered the work. Under the Supreme Court’s Fourth Estate decision, a pending application is not enough. The ordinary registration process takes months, which is why expedited handling exists for suits that cannot wait. Timing matters even more than the filing requirement. Statutory damages and attorneys’ fees are available only if the work was registered before the infringement began or within three months of first publication. Infringement “commences” at its first act for this purpose. Register late and you are limited to actual damages and profits, which are harder to prove and often smaller. Registration within five years of publication also gives you a presumption that the copyright is valid. Once registration issues, the suit can reach infringement that happened before it. The registration clock is the single cheapest piece of litigation leverage a software company can buy, and most companies never buy it.
The limitations period has its own strategy. Suit must be filed within three years of when a claim accrues. Most courts run that clock from when you discovered or reasonably should have discovered the infringement. Under the Supreme Court’s decision in Warner Chappell v. Nealy, a timely claim supports damages reaching back to the start of the infringement, with no separate three-year cap on recovery. That matters in software cases, where copied code can hide inside a competitor’s product for years. How far the discovery rule itself extends continues to be tested in the courts, so treat the long damages tail as leverage rather than a guarantee.
Copyright law protects the expression in your program, not the ideas behind it. The statute itself excludes ideas, procedures, processes, systems, and methods of operation from protection (17 U.S.C. § 102(b)). That territory belongs to patents and trade secrets, if it is protected at all. Computer programs, meaning the sets of instructions that make a computer do something, are protected as literary works in both source code and object code form. Protection can extend past the literal code to non-literal elements like the program’s structure, sequence, and organization, but only where those reflect creative choices rather than function.
Most courts follow the abstraction-filtration-comparison test from Computer Associates v. Altai as their framework when they compare two computer programs to determine whether they are substantially similar. They adapt it to the technology at hand. The court breaks the plaintiff’s program into levels of abstraction, then filters out everything copyright does not protect:
What survives filtration is the protected core, and only that gets compared to the defendant’s program for substantial similarity. Two practical consequences follow. First, these cases are decided by expert analysis of code, not by eyeballing the products. Similar-looking screens do not by themselves establish copying of the code underneath, since two entirely different programs can produce the same output. (A copied user interface can support its own claim; it is a different claim from copied code.) Second, the filtration step is where defendants win. A defendant that documents the open source, the standard techniques, and the externally dictated elements in the plaintiff’s program can shrink the protected core dramatically. A plaintiff that cannot answer that showing, by identifying protected expression that survives it, can lose the case outright.
The strongest cases involve literal copying: identical or near-identical source code, shared comments, replicated bugs. Those fingerprints are hard to explain away, which is why forensic comparison of repositories comes early in any serious case.
Non-literal claims, aimed at a program’s architecture or its look and feel, are harder. Courts in some circuits have refused protection to elements like menu command hierarchies as methods of operation, though that rule is circuit-dependent rather than national. The law on interfaces remains unsettled. In Google v. Oracle, the Supreme Court held Google’s copying of API declarations was fair use without deciding whether APIs are copyrightable at all. If your claim depends on structure rather than code, expect the filtration fight to be the whole case.
A software copyright defendant has more tools than most commercial defendants:
Plaintiffs should also choose their supporting state-law claims carefully. Federal copyright law preempts state claims that just repackage copying, and courts regularly dismiss unfair-competition and unjust-enrichment counts on that basis. Claims with a qualitatively different element survive. Breach of contract does, because the promise is the extra element. So does trade secret misappropriation, because secrecy is an element trade secret law requires and copyright law does not. The federal Defend Trade Secrets Act also supplies a parallel federal claim that avoids the preemption question entirely. That is one reason trade secret counts travel with most software copyright suits. When the same facts involve confidential source code or algorithms, the trade secret claim is often the stronger half of the case. That claim is covered on the Software Trade Secret Disputes page.
If your software uses license keys, dongles, authentication, or other technical measures to control access, the Digital Millennium Copyright Act adds a separate claim. It reaches anyone who circumvents them without authorization or traffics in circumvention tools (17 U.S.C. § 1201). DMCA claims carry their own statutory damages of $200 to $2,500 per act of circumvention, so they scale quickly against a defendant who hacked a protection scheme hundreds of times.
The courts have policed the boundary, though not uniformly. The Federal Circuit requires a circumvention claim to bear a reasonable relationship to protecting copyrighted work. That is how Chamberlain v. Skylink turned back a garage-door opener manufacturer’s DMCA claim over competing remotes. The Ninth Circuit, by contrast, treats circumvention as a standalone violation with no infringement nexus required. Lexmark v. Static Control rejected a printer-cartridge claim on a different ground: a measure that does not effectively control access to the program protects nothing. The Copyright Office’s triennial exemptions add carve-outs of their own, which have included good-faith security research and repair of software-enabled devices. The through-line holds everywhere: the DMCA protects your software’s locks, but it is not a general-purpose weapon against competition. One more front: courts have divided on whether scraping code while stripping its license and attribution information violates the DMCA’s copyright-management-information provisions. The AI training cases have put that issue squarely in play.
A successful plaintiff can obtain injunctions, impoundment of infringing copies, and either actual damages plus the infringer’s profits or, with timely registration, statutory damages, plus attorneys’ fees in the court’s discretion. Injunctions are not automatic: since the Supreme Court’s eBay decision, irreparable harm has to be proven, not presumed. For defendants, exposure is shaped by the same registration timing, by the filtration analysis, and by how fast contamination is contained.
Both sides’ outcomes trace back to decisions made in the first weeks: preserving repositories and version history, getting a qualified expert into the code to identify what was copied and which components were dictated by function, checking the registration record, and reading every license that touches the disputed software to determine what was actually permitted. Courts have seen defendants go as far as decompiling and recreating a product to dodge its license. The paper trail was what told the story.
If you suspect your code has been copied, or you have been accused, the technical record is aging right now. That is worth a conversation with counsel who can read the code as fluently as the complaint, before the dispute defines itself without you.
Yes. Copyright attaches the moment original code is fixed in a tangible medium, in source or object form, with no filing required. But you cannot sue until the Copyright Office has actually registered the work, not merely received your application, and statutory damages and attorneys’ fees require registration before the infringement began or within three months of first publication. Expedited handling exists for emergencies. Automatic protection without timely registration is protection without teeth.
Possibly. Copyright can reach a program’s non-literal elements, like its structure, sequence, and organization, where those reflect creative choices. Courts apply the abstraction-filtration-comparison test to strip out ideas, efficiency-driven code, externally dictated elements, and public-domain material, then compare what remains for substantial similarity. Whether enough survives filtration is the central question, and expert code analysis answers it.
Often, yes. Courts have held that disassembling a program to reach its unprotected ideas and functional elements, for a legitimate purpose like building interoperable products, is fair use. But the protection has limits: copying expressive code into the final product is still infringement, circumventing technical protection measures can violate the DMCA, and contract clauses prohibiting reverse engineering have been enforced by some courts. The full answer depends on what was copied, how, and what the license says.
Injunctive relief, impoundment of infringing copies, and money: either your actual damages plus the infringer’s profits attributable to the infringement, or statutory damages if you registered in time, plus attorneys’ fees in the court’s discretion. A timely claim is not capped at three years of back damages, so recovery can reach the start of a long-running infringement. If circumvention of technical protections is involved, DMCA statutory damages accrue per violation and can dwarf the underlying copyright numbers.
Only claims with an element the copying itself does not supply. Breach of contract and trade secret misappropriation routinely survive federal preemption. Unfair-competition, misappropriation, and unjust-enrichment claims that simply restate the copying usually get dismissed. Pleading the right companion claims at the start avoids losing half the case on a motion.