by Dennis Crouch
In 2025, the Trump administration is in the process of substantially reducing the federal workforce while also signaling interest in increasing reliance on Artificial Intelligence in federal government operations. In this environment of rapid transformation, where once-unthinkable changes are becoming daily realities, the opportunity for radical reform of government institutions has never, in my lifetime, been more tangible.
Amid this dramatic reshaping of federal operations, I wanted to explore a potential structural reform: allowing patent applicants to obtain pre-filing certification of their applications through rigorous outside examination equivalent to USPTO standards. This certification could then facilitate expedited examination at the USPTO, similar to how the Patent Prosecution Highway (PPH) program operates for international applications, but with important distinctions that could make it more effective. The U.S. patent examination system faces persistent challenges of long pendency times, substantial unpredictability, variable quality despite the USPTO’s continued efforts at improvement, and high costs. Although not without potential serious concerns, the proposal could address many of these elements.
The core idea is straightforward: Before filing, applicants would submit their application to a certification entity that would conduct a thorough prior art search and patentability analysis. There may be a certification process – that requires the entity to meet quality metrics parallel to those used by the USPTO internally. Applications certified as meeting patentability requirements would then be eligible for highly streamlined USPTO registration process.
Although AI tools would not be a required element of this process, I expect that they would be integral to the process of both certification entities and patent applicants seeking to navigate this new system effectively. Modern AI systems, particularly those built on large language models and specialized patent examination databases, could dramatically enhance the preliminary examination process by identifying relevant prior art, conducting an analysis for both anticipation and obviousness, suggesting claim amendments, and flagging potential §112 issues. Leaving quality aside for a moment, the benefit of the AI system is its potential cost effectiveness and its timeliness. In particular, this process could compress the examination process into a single day rather than the multi-year scenario that 95% of applicants face.
While this proposal is primarily a thought experiment, its greatest value may be in demonstrating how AI-driven preliminary analysis and immediate feedback could be implemented within the USPTO’s existing framework to improve patent examination.
Learning from the Patent Prosecution Highway
For many folks, the idea of outsourcing examination may seem like a very crazy idea. I do want to pause for a moment to recognize that it is already happening for tens of thousands of applications every year under the Patent Prosecution Highway. The PPH program is a collaborative framework between patent offices worldwide that allows applicants to leverage positive examination results from one jurisdiction to expedite examination in another. When we look at the data, PPH applications show remarkably better outcomes across key metrics – substantially higher allowance rates; fewer office actions per case; and dramatically shortened pendency periods.
While PPH applications already show impressive outcomes, a significant portion of PPH rejections stem from the inherent friction of patent applications crossing jurisdictional boundaries. Applications initially drafted for the Chinese patent system, for instance, often face rejection not because of systemic differences in claiming practices, written description requirements, and other patentability standards. These applications can require substantial reformation to meet USPTO requirements, even when the underlying invention is clearly patentable. In contrast, the proposed external certification system generally be designed for the the U.S. patent framework. Applications would be pre-examined specifically for USPTO standards based upon domestic patentability requirements. This alignment of drafting, examination, and legal standards should produce even higher allowance rates than we see in the PPH program — assuming that we adopted a straight conversion. The certification entities would essentially be conducting a dry run of USPTO examination, without the complications of international translation – both literal and legal.
The PPH system includes some important elements – such as a detailed claim mapping requirement, preservation of USPTO’s final examination authority, and ongoing quality monitoring between participating offices. The basic point is that we already have one working model where USPTO examiners regularly rely upon external examination work product.
But we should be careful about drawing too direct a parallel. The PPH program operates within an established framework of trust between national patent offices, each operating under well-understood governmental oversight. My proposal for private certification entities would require building new trust frameworks and accountability mechanisms. The PPH also maintains full USPTO examination authority, just expedited, while my proposal envisions a more substantial delegation of examination functions. These are important differences that could create insurmountable roadblocks to implementation. Still, I recall the strong push-back against PPH 15 years ago as it was moving forward — arguments that the work of examiners in other countries simply could not be trusted. But instead, the PPH program has demonstrated that the agency can successfully incorporate external examination and offers an important roadmap.
The Role of Artificial Intelligence
While human examiners could certainly conduct the pre-filing examinations, the scalability and cost-effectiveness of this proposal likely depends on the integration of artificial intelligence tools. The key would be establishing appropriate validation frameworks to ensure AI-driven examination meets quality standards. Pre-filing certification entities would need to demonstrate that their automated systems achieve acceptable performance across the USPTO’s quality metrics.
Advanced AI systems could significantly enhance the efficiency and effectiveness of this pre-filing certification process. These tools could analyze draft patent applications against the prior art, suggesting claim language modifications to overcome potential §102 and §103 rejections. The AI systems could perform semantic analysis to identify terms requiring clarification under §112, while simultaneously cross-referencing the specification to ensure adequate written description and enablement support. One solution then would be to modify claim language, and another would be to suggest additions to the written description. By leveraging large language models trained on successful and unsuccessful patent prosecution, the AI tool could propose alternative claim language that maintain broad coverage while avoiding identified prior art. As part of the process, there will likely need to be a human in the loop on the certification side, but the hope is that individual can be provided with an evidence-supported report for approval or further inquiry.
Addressing the Secret Prior Art Problem
One potential challenge for pre-filing certification involves so-called “secret prior art” under 35 U.S.C. § 102(a)(2) – applications filed before but published after the examined application’s filing date. However, this concern could be substantially mitigated by following the European Patent Office’s approach of excluding such references from obviousness considerations. This change would align U.S. practice with other major patent offices and eliminate a significant source of uncertainty in pre-filing patentability determinations. It would also benefit applicants conducting traditional pre-filing searches by removing the unpredictable element of yet-unpublished applications.
The Current Quality Framework:
The USPTO already has an extensive process for measuring and ensuring patent examination quality. For instance, the Office of Patent Quality Assurance (OPQA) conducts reviews that focus proper claim interpretation, search completeness, and the accuracy of patentability determinations under 35 U.S.C. §§ 101, 102, 103, and 112. This approach is a good starting point for Any pre-filing certification system would need to demonstrate comparable performance across these dimensions. The USPTO Patent Quality Assurance program adds additional layers of review, focusing on consistency across art units and tracking trends in examination quality over time. These metrics provide a framework for validating the work of certification entities. One difficulty here is that a substantial amount of quality assurance comes through individual review and management by primary examiners and supervising patent examiners (SPEs).
Establishing a Pilot Program
Given the magnitude of changes proposed and the legitimate concerns around quality assurance, the most prudent path forward would be through a pilot program to allow the USPTO to evaluate the potential of AI-facilitated examination. The pilot could begin with a single technology center, where AI tools have already demonstrated particular promise in prior art searching and analysis. Working with one or more US-based AI providers who have shown their focus on patent examination, the USPTO could establish a controlled testing program that involves working with volunteer patent applicants.
Implementation Considerations
Although I’ve presented this proposal primarily as a thought experiment, we should carefully consider the practical hurdles to implementation. And I want to be clear – this is not meant to denigrate our corps of patent examiners who work diligently to ensure the quality of U.S. patents. Rather, this is about exploring structural reforms that could potentially make their work more effective while improving the system for applicants.
The most immediate challenge involves quality assurance and oversight. While we can imagine various certification frameworks, the reality of implementing them without compromising examination integrity presents really difficult challenges. The USPTO’s current quality framework relies heavily on the judgment examiners, including oversight of primary examiners and SPEs – human expertise that would be difficult to replicate in private entities – especially in AI tools. We would need to develop new mechanisms for ensuring consistent quality across multiple certification entities while preventing a “race to the bottom” where entities compete for business by offering lenient examination standards.
Perhaps the most serious concern involves the potential for corruption or improper influence. We would need careful thought about conflict of interest and oversight mechanisms. This includes the very serious concern regarding management of confidentiality of patent applications during the certification process.
In many ways, this is a market based proposal, and so the economics of the system would need to work so that the process is truly cost effective with improved results in terms of both quality and timeliness.
I expect the greatest value of this thought experiment will not be in its literal implementation, but in how it shows possibilities for internal USPTO reform (and provides pressure to move in that direction). Rather than outsourcing examination functions, the core concepts here – particularly the integration of AI-driven preliminary analysis and immediate feedback mechanisms – could be deployed within the USPTO’s existing framework. The Office could develop tools that provide applicants with automated pre-filing guidance on potential prior art conflicts, §112 issues, and claim drafting suggestions, while maintaining traditional examiner oversight. This would preserve the essential human expertise and institutional safeguards of the current system while leveraging technological advances to improve quality and efficiency. Such an approach would achieve many of the benefits discussed above without introducing the significant challenges of external certification.
I have extensively reviewed and exposed the poor patent examination quality of some PPH filed applications.
Not all countries of first filing operate under the same quality framework.
Many Chinese-originating patent applications that are initially filed and examined in the CNIPA are considered to be of poor quality. In 2021, CNIPA itself identified about 15% of all Chinese patent applications as irregular (or abnormal) with a withdrawal rate of 93%. The Special 301 Report issued by the United States Trade Representative in April 2023 says that “large quantities of poor-quality patents continue to be granted” by the CNIPA.
Some Chinese-originating applications rush through the USPTO on the Patent Prosecution Highway, leading to hasty allowances that essentially rubberstamp CNIPA decisions.
From my snapshot analysis of of 300 PPH application, those filed by China, (typically first examined by the CNIPA) are more rapidly allowed – often in a first action or Quayle action – by the USPTO than those filed by entities and first examined in other patent offices.
Letting CNIPA patentability decisions color the US patent examination process may be more efficient but is it fair to all inventors?
Thanks for compiling some interesting out-of-the-box ideas, Professor Crouch.
As a former USPTO Quality Assurance Specialist, I see shortcomings with a proposed requirement that any pre-filing certificate system “demonstrate comparable performance” as the current quality framework.
The USPTO’s Office of Patent Quality Assurance survey results for FY2023 show that
24.4% of the 112(a) enablement rejections were improper
18.3% of the 112(a) written description rejections were improper
15.4% of the 112(b) indefiniteness rejections were improper and
25.2% of the 112(d) rejections were improper.
Using this low quality baseline for a pre-filing certificate program, we are fast approaching the scenario where it’d be just as efficient to let patent examiners flip a coin.
After all, that’s what it’s all about these days, isn’t it? Efficiency.
Not to sound cynical, but I assumed that this was the point of hollowing out federal functions, so that the work could be outsourced to private sector companies owned by hand- selected individuals.
Agree that “We would need careful thought about conflict of interest and oversight mechanisms. ” We know that AI is trained on published and issued patents. One concern is the privacy of patent applications. As the Atlantic recognizes, giving access to outside individuals gives a “god mode” view of emerging and valuable tech. link to theatlantic.com
I think that there have been a few stated goals: (1) generally reducing the size of the fed gov’t is designed to reduce the power of the federal gov’t to regulate private industry and also to regulate individual states. (2) there is an idea that current federal workers are more likely to be liberal, or at least proponents of the status quo; it will be easier to operate more radically if those folks are reduced; (3) there may need to be some rehiring, but those can be people hired into a new normal and perhaps folks more supportive of the current administration.