AI Copyright Lawsuits in Federal Court

Home › Guides › AI Copyright Lawsuits in Federal Court

✓ checked 11 October 2026 · sources: CourtListener / RECAP federal dockets and complaints (Free Law Project) · method · caveats · download the data

We found 97 federal copyright dockets, filed since 28 Jun 2023, whose complaints concern the training of AI models, against 19 named defendant groups. 11 were filed in 2023, 15 in 2024, 29 in 2025 and 42 in 2026 to 8 October. The latest, USA Today Co. v. OpenAI Foundation, was filed on 8 Oct 2026.

Read this first. Every complaint is a set of allegations by the plaintiffs. None of the statements quoted here is a finding by a court, and a docket being listed says nothing about who is right. “Closed” means CourtListener records a termination date, which can follow a dismissal, a settlement, a transfer or a consolidation, and the docket data don’t say which.

dockets 97with no termination date 75in S.D.N.Y. or N.D. Cal. 81

Headline numbers

1. 97 dockets, 75 with no termination date and 22 closed. The count rises each year: 11, 15, 29 and 42 so far in 2026.
2. Two courts hold most of them: the Southern District of New York has 31 (32.0%) and the Northern District of California 50 (51.5%). In all, 9 federal courts are involved.
3. Defendants: the most frequent defendant in the case name is OpenAI (23 dockets), then Meta (14), Anthropic (13) and Suno (8). Some defendants are large technology companies sued over an AI product; the complaint text decides inclusion, not the company name.
4. Not every docket is a separate dispute. 15 dockets belong to 7 plaintiff-and-defendant pairs that appear more than once, usually because a case was refiled, transferred or split. We keep each docket because each has its own number and court.

Filings by year and by defendant

What the complaints say

These are quotations from complaints. They are the plaintiffs’ allegations and aren’t findings of fact. The USA Today plaintiffs, in the suit filed on 8 October 2026, state: “the USA TODAY Plaintiffs seek damages in excess of $250 million”. The New York Times Company’s December 2023 complaint against Microsoft and OpenAI says the action “seeks to hold them responsible for the billions of dollars in statutory and actual damages that they owe for the unlawful copying and use” of its works. The authors in Authors Guild v. OpenAI allege that “at the heart of these algorithms is systematic theft on a mass scale”. The defendants dispute such claims in court; we don’t summarise their responses because we haven’t read them.

For the wider legal picture on scraping, see the Scraping Law Hub and our look at how scraping cases end. For what sites say about AI crawlers, see the AI opt-out survey.

All 97 dockets, newest first

Method

Candidates. On 11 October 2026 we searched CourtListener’s RECAP dockets for federal cases with nature of suit 820 (copyright) filed since 1 June 2022 and a case name containing one of 30 company names (AI developers such as OpenAI, Anthropic and Suno, and large technology companies such as Meta, Microsoft, Google and Apple). That returned 181 results, which are 181 distinct dockets when matched on court and docket number. We set aside 14 miscellaneous subpoena matters (docket numbers containing “-mc-”) and downloaded the complaint for 142 of the remaining dockets.

Classification. We downloaded each docket’s first filing (the complaint) from the RECAP archive and converted it to text. A docket is included if the complaint uses a form of “train” at least 15 times and AI vocabulary (“artificial intelligence”, “AI”, “large language model”, “LLM”, “generative”, “machine learning” or “neural network”) at least 10 times. Dockets whose complaint we couldn’t download were excluded. Cases where the complaint is not about training, such as music-licensing disputes with a technology company, fall out under this test.

Status. “Closed” uses CourtListener’s termination date for the docket.

Verification. A separate script converts the complaints again with different settings, reapplies the test with its own code, recomputes every figure and checks each quote against the complaint text.

Caveats

• This is a floor. We only searched copyright cases whose case names contain our list of companies. Suits under other categories, against other defendants or with different names are missing, and so are cases not in the RECAP archive.
• The text test is a rule, not a legal judgement. It can include a dispute that is mainly about something else and miss a case whose complaint is short or in a different form. The multidistrict case In re OpenAI Copyright Infringement Litigation (S.D.N.Y. 1:25-md-3143) is a consolidation docket with no complaint of its own, so it isn’t counted.
• Docket counts aren’t counts of disputes (see above) or of plaintiffs; one docket can have many plaintiffs.
• Complaint quotations are allegations. We report no outcomes beyond the termination dates on the dockets.

Download the data

• Federal AI-copyright dockets (CSV): court, docket number, case name, dates, defendant group and CourtListener link for all 97 dockets.

Cite as: ProxyPicker, “AI Copyright Lawsuits in Federal Court”, 11 October 2026, https://proxypicker.com/ai-copyright-lawsuits/. Data: CourtListener (Free Law Project), RECAP archive.

Related: Scraping Law Hub · How Scraping Cases End · The AI Opt-Out Standard vs Real robots.txt Files.

Press & data requests: email [email protected] for interviews, the raw data behind this page, or corrections. See also our contact page.