Reuters reported on 12 November that Anthropic and OpenAI had been sued over inventors’ neural-network patents. The court records show two complaints filed that day in federal court in Delaware by Flinn Artificial Intelligence (AI) Technologies, a Texas company, one against Anthropic and one against OpenAI. Each asserts the same 5 patents and demands a jury trial.
The complaints allege that both companies built their businesses “based on the architectures recited in the Asserted Patents”. Those are allegations; no court has ruled on them. Anthropic filed for its IPO on 26 October, and the OpenAI complaint says OpenAI confidentially submitted a draft S-1 in June.
Business Pill · IPO Float
A short explainer on IPO float, the share of a company sold in its listing. It teaches the general idea and does not discuss these lawsuits.
The key insight: As we read the two complaints, the claim is aimed at the attention-based architecture both labs build on, not at one product, and the patents asserted were issued in 2025 from applications claiming 2015 and 2016 priority.
The patents name two inventors, Steven D. Flinn and Naomi Felina Moneypenny, who assigned them to the plaintiff. The complaints say the pair filed applications on 3 August 2015 and 18 January 2016 “describing iterative attention-based neural-network architectures”, and that each asserted patent claims priority to them.
The accused products are broad. Against Anthropic, the complaint lists Claude Chat, Claude Code, Claude Cowork, Claude Design, Claude Enterprise and API access, powered by Claude models from Claude 1 onward. Against OpenAI, it lists ChatGPT, Codex, the API platform and models from GPT-3.5 onward, among others.
The OpenAI complaint cites OpenAI’s own GPT-4 Technical Report and API documentation to say its models are transformer-based. Both complaints say Flinn makes no product covered by the patents and has licensed them to no one, so no marking was required for past damages.
The five asserted patents, from application filed to patent issued, in years after the 3 August 2015 priority application. Dates from the complaints and Google Patents; years are our arithmetic.
The Five Patents, by Date
The oldest application among the five, for the ’404 patent, was filed in January 2023 and issued in February 2025. The other four were filed between November 2024 and January 2025 and issued between June and December 2025, 9.9 to 10.4 years after the August 2015 priority date, by our arithmetic.
The five are U.S. Patent Nos. 12,223,404, 12,327,166, 12,462,197, 12,462,198 and 12,511,580. Dates are from the complaints and Google Patents records; all five share the title “Iterative attention-based neural network training and processing”. The complaints say the inventors’ portfolio comprises “at least eighteen United States patents”.
For scale, the transformer paper, “Attention Is All You Need”, was posted to arXiv on 12 June 2017, about 22 months after the August 2015 priority date, by our arithmetic. The complaints do not cite it. All five patents were issued between February and December 2025.
What Flinn Asks For
Each complaint asks for damages “in no event less than a reasonable royalty”, a finding of willful infringement with enhanced damages “up to three times the amount found”, an ongoing royalty after judgment, interest, costs and attorneys’ fees. Neither names a dollar figure.
The willfulness claim rests on knowledge “since at least the filing and service of this Complaint”. Neither complaint describes a notice letter sent before the suits.
The IPO Angle
Anthropic’s amended S-1, filed 10 November, warns: “We also face patent infringement claims, which may seek substantial damages or injunctive relief limiting our ability to use certain technologies”. It adds that such claims may increasingly be asserted “as our public profile grows”.
The filing’s legal-proceedings section names copyright and government cases, not a patent case, by our text search. It says Anthropic held 40 issued U.S. patents as of 30 September. Our count of its risk factors, IPO map and share-count rebuild set out the rest of the filing.
The complaints open with each company’s growth: the Anthropic one cites a run-rate of about $65 billion by July 2026 and a $965 billion post-money valuation; the OpenAI one cites a run-rate above $40 billion and an $852 billion valuation. Earlier on 12 November we covered a separate suit by nonprofits against both labs.
The Structural Read
As we read the dates, the asserted patents are recent continuations of older applications: filed between 2023 and 2025, issued in 2025, each claiming priority to August 2015 and January 2016 filings, by the complaints’ own account.
As we read the relief sought, the complaints ask for money, not an injunction: a reasonable royalty at minimum, enhanced damages for willfulness and an ongoing royalty after judgment.
Anthropic S-1/A No. 2, risk factors, 10 November 2026
“We also face patent infringement claims, which may seek substantial damages or injunctive relief limiting our ability to use certain technologies”
Three Implications
FOR ANTHROPIC’S IPO The amended S-1 already warns of patent claims; it names no patent case in its legal-proceedings section.
FOR OPENAI The complaint lists ChatGPT, Codex and the API platform among the accused products.
FOR OUR COVERAGE We will report the companies’ responses and the first court filings in both cases.
The framework says: “The transformer is a commodity now; what changes between paradigms is where intelligence accumulates and who can capture it.”
As we read it, these complaints test that premise from the legal side: they allege the attention architecture itself is patented. The framework predates the suits and does not discuss them.
What Is Not Established
No court has ruled on whether the patents are valid or infringed, and no judge was assigned when we read the dockets. Neither company’s response was in the material we read, and Reuters’ full article was not available to us. The complaints give no damages figure. This case is separate from the University of Tennessee Research Foundation’s July patent suit against Anthropic.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
A Texas patent owner sued Anthropic and OpenAI in Delaware on 12 November over 5 patents that claim priority to 2015 and 2016 attention-network applications and were issued in 2025. The complaints allege infringement across Claude and ChatGPT and seek at least a reasonable royalty, up to tripled for wilfulness; both companies are preparing IPOs.
A note on sourcing. This piece reports allegations made in two complaints filed in the U.S. District Court for the District of Delaware on 12 November 2026; they are claims, not findings. Patent dates come from the complaints and Google Patents records; S-1 text from Anthropic’s amended registration statement on EDGAR. Year spans are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
