AI Lawsuits Tracker

AI copyright infringement in these lawsuits refers to AI companies scraping and training their models on copyrighted works — images, text, and code — without a license or compensation to rights holders. The core legal question is whether using copyrighted data to train a machine learning model constitutes infringement of the original creator's exclusive rights, or whether it qualifies as "fair use," a doctrine that permits limited use of copyrighted material for transformative purposes. Courts are still deciding this question, with no definitive ruling yet.

Individual creators can sue AI companies if they can demonstrate their copyrighted work was used without permission in training data. The practical challenge is evidence: tools like Have I Been Trained and Spawning's opt-out databases identify which works appear in known datasets, and artists have documented cases where AI models reproduce their distinctive styles as supporting evidence. If individual litigation isn't affordable, joining an existing class action against companies like Midjourney, Stability AI, or OpenAI is often the more viable path — class action plaintiff attorneys typically work on contingency, meaning no upfront cost to you and a fee only if the case succeeds, which is why joining is usually more accessible than hiring your own litigator. Check the cases in this tracker for plaintiff attorney contact information.

As of mid-2026, no major AI copyright lawsuit has reached a final plaintiff victory at the merits level — most cases are still working through discovery, motions to dismiss, or appeals. A few cases have reached partial settlement terms, but none represent a definitive ruling that AI training constitutes copyright infringement. The closest outcomes so far involve cases where specific AI outputs were found to directly reproduce copyrighted work, rather than the broader question of whether training data use itself is infringing.

The legal status of AI training on copyrighted work remains unsettled in the US. Courts are actively deciding whether training on copyrighted data constitutes fair use — a question that has no clear precedent because AI training is a novel process with no direct historical analog. Multiple cases are working through federal courts simultaneously, and the outcomes will likely conflict until the Supreme Court or Congress establishes a definitive standard. Several countries (EU, UK) have created explicit AI training exceptions or opt-out rights that the US currently lacks.

Individual artists can join existing class action lawsuits against AI companies — cases like Andersen v. Stability AI, Tremblay v. OpenAI, and similar suits are structured to include any copyright holder whose work appears in the relevant training datasets. Check the case dockets via PACER or contact the plaintiff attorneys listed in the tracked cases. Class action participation typically requires submitting a claim form and demonstrating your work appears in the training data, which opt-out databases and dataset auditing tools can help establish.

AI companies argue their training process constitutes "fair use" — a legal doctrine that permits certain uses of copyrighted work without permission when the use is transformative, non-commercial, or doesn't harm the original market. Their core argument is that training is transformative: the AI learns patterns from copyrighted works without reproducing them, analogous to how a human artist learns by studying others' work. Plaintiffs counter that this comparison fails because AI companies build commercial products at massive scale on creators' work without any compensation, and AI output can directly substitute for and compete with the original creator's work.

Yes — under U.S. copyright law you generally must register your work with the U.S. Copyright Office before you can file an infringement lawsuit in federal court, a requirement the Supreme Court confirmed in Fourth Estate Public Benefit Corp. v. Wall-Street.com (2019). Registration also determines what damages are available: registering before the infringement began (or within the early-publication grace period) preserves your right to statutory damages up to $150,000 per work plus attorney's fees, while late registration typically limits you to actual damages, which are harder to prove and often smaller. If your work isn't registered yet, you can still register now — the standard process usually takes several months, with an expedited option for a higher fee — and pursue AI companies for future infringement even without statutory damages for past use. Unregistered creators can sometimes still participate in a class action settlement without individually filing suit, so check the specific case terms in this tracker.