Disclosure or it didn't happen — and a bank disclosed a model patent. Capital One's grant US12443802B2 (“Large language model summarization,” issued 2025-10-14) is a reminder that the serious AI builders aren't only the labs. Assigned to Capital One Services, LLC and classified CPC G06F 40/40 and G06N 3/045, it covers summarizing documents with a large language model.
The business case is document density. Finance runs on long documents — disclosures, contracts, statements, regulatory filings — and summarizing them accurately is genuinely valuable and genuinely hard to do safely. A bank building and patenting its own summarization IP suggests it is treating AI as core infrastructure, not a vendor add-on.
Capital One's filings discuss technology and AI investment as part of its operating strategy without isolating technique-level economics, which is normal for a financial institution. The grant is the granular record under that strategy: dated 2025, owned, aimed at a high-value, regulated workflow.
House discipline: a grant proves invention and ownership, not a revenue figure, and we attribute none. It also doesn't establish deployment in a customer-facing product. What it documents is that a major regulated bank holds dated LLM IP — a signal about who is actually building AI capability.
For the markets reader, the reusable point is to widen the lens beyond the obvious AI names. When a bank holds a 2025 LLM-summarization grant, it tells you AI capability is diffusing into regulated incumbents — and that the build-versus-buy line in financial-services AI is further toward “build” than the headlines suggest.