Flinn sues Anthropic and OpenAI over five AI patents | Magica
Flinn AI Technologies sues Anthropic and OpenAI over neural-network patents
Editorial Team
••📖7 min read
Flinn AI Technologies alleges that Claude and ChatGPT infringe five neural-network patents. The suits arrive during Anthropic’s IPO roadshow, but its complaint acknowledges that key evidence about Claude’s internal operations must come through discovery.
Flinn AI Technologies filed separate Delaware lawsuits alleging that Claude and ChatGPT infringe five neural-network patents.
The patents were issued in 2025 and claim priority to applications filed in 2015 and 2016; attention research predates those applications.
The Anthropic complaint seeks royalties and damages while acknowledging gaps in public knowledge of Claude’s internal operation.
Flinn Artificial Intelligence (AI) Technologies, LLC, a Texas patent-holding company, sued Anthropic, the developer of the Claude AI models and assistant, and OpenAI, the company behind ChatGPT, in Delaware federal court on November 12, 2026. It alleges that both companies infringe five patents covering neural-network operations used in reasoning, search, memory, coding and other capabilities.
The lawsuits seek payment for technology the companies sell across their AI businesses. For Anthropic, the challenge arrives during its initial public offering roadshow. Yet Flinn’s own complaint acknowledges that Claude’s internal operations are not fully public: proving its allegations will require more than showing that the products offer similar capabilities.
A claim on the machinery behind Claude and ChatGPT
The defendants are Anthropic PBC and OpenAI Opco LLC, in cases numbered 1:26-cv-01454 and 1:26-cv-01455, respectively, according to the report on the complaints. Neither company immediately responded to requests for comment for that report.
The dispute adds an architectural challenge to companies already facing multiple lawsuits over copyrighted training material. Patent suits against the two have been relatively rare, the report notes. Here, the allegation concerns how the systems process information and generate or act on outputs. It is not a finding that either company copied an invention or infringed a valid patent.
A 2,000-cubic-metre tank and mobile pumping capacity address northern Hong Kong flooding, while AI tools help position crews and detect rising water. Official records distinguish established monitoring from three-site forecasting trials, with costs disclosed but comparative performance results absent.
Editorial Team
The detailed Anthropic complaint reaches beyond a chatbot interface. It names Claude Chat, Claude Code, Claude Cowork, Claude Design, Claude Enterprise, API access and hosted or dedicated deployments. Flinn alleges direct infringement as well as inducing or contributing to infringement through customers, developers and distribution partners.
Flinn connects the accused capabilities to subscriptions, usage charges, enterprise services and cloud distribution. Its complaint points to Anthropic’s model pricing tiers as evidence of the capabilities’ commercial value. That is the plaintiff’s argument for compensation, not a calculation of damages; the amount sought remains unspecified.
Anthropic said in its November 10 roadshow announcement that it planned an offering of 825 million shares at an expected $70–$78 each. About 801.4 million shares would be sold by the company, implying roughly $56.1 billion–$62.5 billion in gross proceeds before fees and any additional underwriter purchases, calculated from the announced share count and price range.
The rest would be sold by existing stockholders, with no proceeds to Anthropic. These are proposed financing terms, not money already raised or a measure of Flinn’s potential recovery.
Co-inventor Steven D. Flinn is founder and chief executive of ManyWorlds, a developer of machine-learning recommendation and adaptive systems, according to the complaint. It describes him as a former chief information officer and strategy executive at Royal Dutch Shell.
The other named inventor, Naomi Felina Moneypenny, is a Microsoft partner group product manager. The complaint describes her as a former Shell business and technology strategist who subsequently led development of ManyWorlds’ Synxi adaptive recommendation engine.
According to the complaint, the pair developed an earlier portfolio involving adaptive recommendations, personalization and information discovery, tracing to applications from 2010 and 2011. ManyWorlds licensed or otherwise transacted parts of that portfolio with major enterprises. The five patents in these lawsuits belong to a later, separate generation of work.
Flinn AI Technologies says the inventors assigned it the asserted patents and associated rights to recover damages. It describes itself as their exclusive owner, says it has granted no third-party licenses, and states that it does not itself make or sell products covered by the claims. Its position in this dispute is therefore that of an owner seeking payment for alleged use of its intellectual property.
The dates require care. The five asserted U.S. patents—12,327,166; 12,223,404; 12,511,580; 12,462,197; and 12,462,198—were issued between February and December 2025. They claim priority to applications filed on August 3, 2015, and January 18, 2016. The earlier dates describe the claimed lineage of the inventions, not when these five patents were granted.
Attention has a research history beyond this lawsuit
Flinn describes its inventions as arrangements in which attention to different representations of information produces probabilities that guide further attention and processing. Other claims address storing representations for later use, comparing vectors to retrieve content, correcting generated answers, selecting code and performing actions.
The chronology does not mean attention-based language processing began with these patents. A neural translation paper first submitted in September 2014 proposed letting a model search for parts of a source sentence relevant to predicting a target word. The Transformer paper first submitted in June 2017 subsequently proposed an architecture based solely on attention mechanisms, dispensing with recurrence and convolutions.
Those papers establish a broader research chronology; they do not resolve whether Flinn’s particular patent claims are valid or whether Claude or ChatGPT practices them. Flinn itself argues that its claims cover specific ordered arrangements and do not preempt attention-based neural-network processing generally. It identifies alternatives it says remain available, including retrieving content by literal term matching instead of comparing neural-network-generated vectors. Its case therefore turns on the claimed sequence of operations, rather than ownership of attention or search in general.
The complaint says the five patents collectively contain more than 1,000 issued claims, but explicitly distinguishes that total from the claims it will ultimately assert under court procedures. The size of the portfolio is not a count of proven infringements.
The evidence still needed and the royalties sought
Flinn’s pleading acknowledges that Claude’s architecture, source code, training procedures and intermediate processing are not fully public. The plaintiff says its allegations rest on public information and accompanying claim charts, and that discovery should confirm and supplement them through technical records and testimony.
The complaint cites Anthropic’s 2023 Claude 2 model card, which describes that model and its predecessors as using a transformer architecture. It also cites the company’s materials on reasoning, memory, retrieval and coding. But Flinn repeatedly alleges the specific internal operations on information and belief, including how later models use attention and how memory feeds into subsequent processing. Those are implementation claims that the requested technical evidence would test.
In the Anthropic case, Flinn demands a jury trial and damages no lower than a reasonable royalty. It also requests an ongoing royalty for infringement continuing after judgment, enhanced damages of up to three times the assessed amount for alleged willfulness, and legal fees and costs. For its willfulness allegation, it points to Anthropic’s knowledge of the patents and allegations from at least the filing and service of the complaint. These are requests and allegations, not awards or findings.
The next questions are which patent claims Flinn will pursue under the court’s procedures, how those claims should be interpreted, and whether technical evidence shows that the accused systems perform the specified operations. Their validity and application to Claude and ChatGPT remain to be tested; the complaints alone do not establish a right to royalties.
Reported server price demands reach 45% above earlier agreements. HPE explicitly permits repricing before shipment, while XTX's disputed Dell order raises a sharper question: what happens when a buyer says it paid extra for a fixed price?
OpenRouter disclosed provider removals after investigations into inference data handling. Separate notices describe forwarded requests, a compromised deployment and stolen prompt samples, exposing different weaknesses behind privacy assurances.
Employee equity at Anthropic’s IPO price is reportedly worth $232 billion, while book rightsholders await a first settlement installment estimated at $2,203.56 per work. Vesting, shared copyrights and founder voting rights complicate the comparison.
Moore Threads is preparing a December 29 reveal for Lushan, a desktop GPU pitched at gaming and local AI inference. Its claimed memory record remains unquantified, and capacity alone will not establish performance or value.
Amazon is in advanced talks to buy Decart, whose software helps AI models run efficiently across chips. The reported deal could strengthen AWS’s Trainium business, with conditional payments and five-year founder commitments making execution central to its value.
The University of California is expanding ChatGPT and Codex access across 10 campuses, with extra charges for advanced models and an unresolved dispute over faculty authority.
Ray-Ban’s US store cuts the Gen 2 Wayfarer by 25% to $284.25. Buyers still face separate prescription-lens prices, a cheaper Gen 1 option and newer glasses outside Meta’s frame sale.
Ubisoft has made the team behind its Teammates experiment a dedicated subsidiary targeting AI characters with memory and outside game creators. Earlier playtests show both progress and limitations, while commercial plans remain undisclosed amid group cost cuts and refinancing.
Neuralink has shown a trial participant controlling GTA VI through intended movements. Earlier gaming use and company-reported trial results put the demonstration in context, while its reliability across users remains unmeasured in the reports.
SoftBank founder Masayoshi Son reportedly explored financing a $1 trillion acquisition of memory-chip maker SK hynix, but the company says the deal is no longer under active consideration. The target brings surging earnings, existing shareholders and customer commitments that complicate any simple equation between ownership and access to AI-memory supply.
Nvidia reported $115.8 billion in quarterly revenue, but its shares fell 14% Friday amid a wider technology selloff, a leadership transition and a reported accounting investigation. Strong cash generation offers a counterweight to concerns about the durability of AI spending.
The Nasdaq 100 fell 5.6% on November 20 and closed 19.8% below its record after an intraday bear-market breach. Nvidia’s revenue surge and Anthropic’s huge IPO did not halt the rout; infrastructure bills and resilience elsewhere in the market complicate the picture.
Reporting traces how Megaspeed gained access to Nvidia chips through an unlisted Inspur subsidiary and alleges links to a Cambodian smuggling intermediary. Company denials, differing customs totals and an unresolved criminal investigation complicate the account.
Together AI says an intruder downloaded stored samples of June and July API prompts and searched them for credentials. Zero data retention is now the default for new accounts, but the number of affected customers remains undisclosed.
Users describe grief and physical distress when AI relationships end, whether they choose to leave or the service changes. Research on attachment and farewell tactics raises questions about how providers measure engagement, manage updates and protect intimate data.
UN-backed talks agreed on two more years of work toward an autonomous-weapons instrument, without committing to a treaty. Consensus kept major military powers involved while allowing opponents to weaken the mandate.
Tesla is sending FSD v15 to its first customers, adding a visualization of planned driving and changes to navigation and hazard detection. The software remains supervised, and its claimed 25-fold intervention improvement lacks a disclosed baseline.
OpenAI reports sharply lower reasoning-extraction rates in internal GPT-6.2 tests and backs sanctions against illicit foreign distillation. Its own research and rival Anthropic’s experience show why protecting hidden reasoning leaves capability transfer and customer costs unresolved.
Alberta proposes 600-metre setbacks for large data centres and expanded utility oversight, but exemptions and rules still to be written will determine how much protection communities receive.