USPTO Data Failure Hid 1.4 Million Patent Assignment Records – Including 450,000 Real-Ownership Transfers

Roughly 1.4 million patent assignment records held by the U.S. Patent and Trademark Office sat outside public view for years, according to a long-read investigation published by IAM Media on May 13, 2026. When data hidden during system maintenance was restored, researchers found large numbers of newly visible records—including about 450,000 non-provisional, real-ownership transfers—that should always have been public under U.S. patent law. The episode exposes a basic gap in the transparency of patent ownership data on which licensing, litigation, and economic research routinely depend.

U.S. patent assignment practice rests on 35 U.S.C. §261, which lets patent owners record transfers with the USPTO to perfect their rights against third parties. The resulting register is not just an administrative archive: it is the starting point for identifying the real owner of a patent, evaluating licensing rights, deciding who has standing to sue, performing IP diligence on a public company, and tracking non-practicing-entity activity. Missing records distort all of these uses.

The IAM piece frames the data gap not as a hundred-thousand-record anomaly but as a 1.4-million-record shortfall, with the 450,000 non-provisional, real-ownership transfers as the most consequential subset. Topics tagged on the article include Bank of America, Comerica, Credit Suisse, Dow Chemical, FuboTV, JPMorgan Chase, US Bancorp, and Wells Fargo—suggesting that banks’ patent holdings and transfers, often perceived as a small slice of the IP market, may have been systematically under-reported. The piece also flags non-practicing-entity transactions, hinting that aggregator and monetization activity may look different once the previously hidden tranche is folded into the public record.

On October 20, 2025, the USPTO retired two legacy assignment search applications and consolidated them into a new Assignment Center. The data anomaly may stem either from the migration itself—non-publication flags carried over incorrectly into the new environment—or from earlier mis-flagging that became visible only after the maintenance and restore cycle. The Office has previously disclosed unrelated incidents involving inadvertent exposure of protected application data in Patent Center, underscoring the cross-system data-governance challenge it faces.

The downstream impact is wide. Valuation and licensing depend on accurate chain-of-title. Standing analyses in patent litigation rely on a verifiable trail of ownership. Antitrust and competition reviews of IP-heavy industries use assignment data to map concentration and to track aggregations of patents into NPEs. Economic and legal research that treats USPTO assignment records as ground truth must now be re-examined in light of a previously invisible population of transactions.

Research communities have already started revisiting datasets built on public assignment records. If 450,000 real-ownership transfers were missing for an extended period, prior estimates of NPE activity, of corporate IP holdings, of patent-market liquidity, and of cross-industry patent flows may all need upward revision. The visibility of bank-sector patent holdings is a particularly sensitive variable, given prior literature treating financial institutions as relatively minor patent owners.

The Office now faces a structured set of questions. How complete is the restored data? Have past licensing deals, settlements, or court rulings been based on incomplete records that would have looked different had the hidden tranche been visible? What safeguards prevent a recurrence as the Office continues to consolidate and modernize its data platforms? These questions are likely to surface in oversight hearings on Capitol Hill, where patent policy increasingly intersects with broader debates about data transparency and the integrity of public registries.

There is a wider lesson about open data and migration governance. The USPTO has invested heavily in its Open Data Portal, and AI-driven patent analytics increasingly use that data as a substrate. A long gap in the underlying record threatens the accuracy of every analysis, commercial product, and policy paper built on top of it. How the Office explains the gap, and what corrective measures it commits to, will shape the credibility of U.S. patent information infrastructure for years to come.

Sources:
– IAM Media, “USPTO data failure hid 1.4 million patent assignment records for years,” May 13, 2026 (Angela Morris)
– USPTO, “USPTO enhances assignment records search process” (Assignment Center launch, October 20, 2025)
– 35 U.S.C. §261 (Recording of assignments)