The Invention Secrecy Act of 1951, codified at 35 U.S.C. §§ 181–188, authorizes the Commissioner of Patents to issue a secrecy order when publication of a patent application “would be detrimental to the national security.” The order withholds grant of the patent and restricts disclosure by the inventor. The statute requires the agency to report annually to the congressional armed services committees on the number of orders in effect, new orders issued, and orders rescinded. Those summary figures are among the few consistent public metrics available.

According to the most recent Invention Secrecy Activity reports posted by the U.S. Patent and Trademark Office and compiled by independent trackers such as the Federation of American Scientists, several thousand secrecy orders remain in force in a typical year. New orders continue to be imposed even as older ones are reviewed and sometimes lifted. The reports do not identify the inventions, the applicants, or the sponsoring defense or intelligence components that requested the withholdings. The record shows only aggregate counts and, in some years, a breakdown by whether the application originated with a government contractor or a private inventor.

USPTO Freedom of Information Act annual reports, submitted to the Department of Justice under 5 U.S.C. § 552(e), describe a different slice of the same bureaucracy. Those reports tally requests received, processed, and denied; they list the statutory exemptions invoked, including Exemption 3 for records specifically exempted by other statutes and Exemption 1 for classified national-security information. The FOIA reports do not, however, cross-walk those denials to individual patent applications placed under secrecy orders. A requester seeking a pending application that is subject to an order would typically be told that the record is withheld under the Invention Secrecy Act itself, a withholding that appears in FOIA statistics only as an Exemption 3 citation, if it appears at all.

Officials have said in public testimony and in the agency’s own FOIA handbooks that patent applications remain confidential until publication or issuance under 35 U.S.C. § 122, and that a secrecy order adds a further layer. The FOIA reports therefore reflect processing of other USPTO records—assignment files, examiner notes on published applications, trademark files, and administrative correspondence—more than they reflect the secrecy-order docket. The annual FOIA narrative sections discuss backlog reduction, fee collections, and average processing times. They do not list the number of secrecy-order files reviewed or the number of times a FOIA requester was directed to the Invention Secrecy Act rather than to a standard exemption.

The Government Accountability Office and the USPTO Office of Inspector General have, in past reviews of patent operations, noted the dual reporting streams without reconciling them. GAO work on patent pendency and on interagency coordination with the Department of Defense and the intelligence community has treated secrecy orders as a national-security process rather than a FOIA process. Inspector General audits of FOIA compliance at the Commerce Department, of which USPTO is a bureau, have focused on response times and exemption use, not on whether secrecy-order statistics belong in the FOIA annual report.

Public dockets illustrate the practical gap. When inventors or their counsel have challenged secrecy orders in the Court of Federal Claims or in district court under the compensation provisions of 35 U.S.C. § 183, the litigation files are often sealed or heavily redacted. Those cases do not appear as line items in the USPTO FOIA report. Likewise, congressional notifications required by the Act are transmitted to the armed services committees; they are not appended to the FOIA report that goes to the Justice Department and is posted on the USPTO website.

Budget justifications submitted by USPTO to Congress list funding for patent examination, information technology, and “national security” liaison functions. They do not isolate the cost of maintaining the secrecy-order program or of reviewing orders for possible rescission. The FOIA reports, for their part, itemize FOIA personnel and contractor costs. The two documents therefore describe overlapping but non-identical activities: one counts withheld inventions; the other counts processed information requests.

Independent researchers who file FOIA requests for lists of secrecy-order serial numbers or for the interagency correspondence that triggers an order routinely receive denials citing the Act and Exemption 3. Those denials increment the FOIA statistics. They do not increment, or decrement, the secrecy-order census reported to Congress. The result, according to the public record, is two official ledgers that both concern withheld technical information yet remain unlinked in the annual publications.

The statute itself contains a review mechanism: secrecy orders are supposed to be reexamined periodically, and inventors may petition for rescission. The Invention Secrecy Activity reports record the outcomes of those reviews in aggregate. FOIA annual reports record neither the petitions nor the reviews. A member of the public seeking to understand how many applications moved from secret to publishable status in a given fiscal year must consult the secrecy-order report, not the FOIA report.

Commerce Department and USPTO officials have publicly described the FOIA program as a transparency tool for agency operations other than unissued patent applications. They have described the Invention Secrecy Act as a separate national-security statute administered in coordination with defense and intelligence agencies. Both characterizations appear in hearing records and in the agencies’ own published guidance. Neither document set claims that the FOIA annual report is a substitute for the secrecy-order census, nor that the census is a substitute for FOIA accounting.

The numerical discrepancy is therefore not an allegation of under-counting so much as a structural feature of two reporting regimes enacted at different times for different audiences. One regime answers to the armed services committees with a head-count of suppressed patents. The other answers to the Justice Department with a head-count of processed requests and invoked exemptions. The applications themselves remain, by design of the 1951 Act, outside both public lists.