On October 8, 2025, the United States Patent and Trademark Office (“USPTO”) posted an official notice announcing a new, voluntary Automated Search Pilot Program. If an applicant applies to and is accepted to the Program, it will provide results from a machine-assisted prior art search generated by the USPTO’s internal artificial intelligence (“AI”) tools shortly after an application is filed. In simple terms, this means that applicants can get an early look at prior patents and publications that the USPTO’s algorithms consider most relevant to an invention. This early feedback can be valuable in several ways: it can help applicants refine their claims before examination begins, decide whether to pursue related applications, or assess whether significant prior art might affect the likelihood of obtaining a patent. If the Program is successful, it may allow inventors and applications to improve patent strategy decision-making, helping applicants move forward with greater clarity and confidence in the strength and direction of their patent filings.

Ultimately, the report generated by the internal AI tool will leverage a patent application’s specification to identify up to 10 relevant documents from U.S. patents, pre-grant publications, and foreign prior art. As the report will not be a statutory “notification” under 35 U.S.C. 132, applicants will not be required to respond, although they may choose to submit a preliminary amendment, seek deferral of examination, or abandon the application to claim a fee refund. Moreover, USPTO Examiners may consider the documents listed in the report during examination but are not obligated to list them in their search report unless relying on them.

The USPTO frames the Program as an experiment to test whether earlier disclosure of potential prior art helps applicants shape prosecution strategies, improves examination quality, and is scalable from an operational standpoint. During and after the Program, the USPTO will collect data, solicit feedback, and decide how or whether to expand it.