There is currently a major policy challenge to America’s leadership in artificial intelligence (AI) flying under the radar: patent policy.
While it’s less discussed than other AI issues at the moment, the United States Patent and Trademark Office (USPTO) has proposed substantial changes to how baseless patents can be challenged and reviewed that risk slowing AI growth.
The US patent system plays a critical role in driving innovation. When it works well, it rewards truly novel inventions and encourages companies to invest in new ideas, take risks, and bring better technologies to market. But when the patent system falters, it can raise costs, reward frivolous litigation, and slow progress, especially in fast-moving fields like software and AI.
The USPTO’s recently proposed rules — which would make it significantly harder to challenge questionable patents after they are granted — shape who can build, improve, and deploy new technologies. That has a profound impact on how AI is developed and whether competition thrives or stalls.
Patents Are Hard to Get Right
Patent examiners review hundreds of thousands of applications each year, many involving complex technologies and dense technical claims. They must make difficult judgments about whether to grant a patent, often under significant time pressure and using only the applicant’s perspective. Given those realities, some patents are issued that should never have been granted, or that have claims broader than what was really invented.
Because it costs defendants millions of dollars to successfully prove that a patent is invalid, the incentive to settle is high — and bad actors take advantage of this. By using weak patents to threaten litigation, a bad actor can extract large settlements from multiple victims using the same invalid patent.
Why Post-Grant Review Exists
Congress specifically acted to address these issues when it created post-grant review procedures as part of the America Invents Act. One of the most important tools created by that law is inter partes review (IPR), which allows the USPTO to take a second look at a patent if someone brings sufficient evidence that a mistake was made.
The logic is straightforward. If new information shows that the USPTO likely made a mistake, the expert agency that issued the patent should have the authority — and responsibility — to correct it. The process is far faster and less expensive than going to court.
USPTO Has Recently Undercut the IPR Process
In recent months, the USPTO has adopted new policies and proposed rule changes that make it virtually impossible to use IPRs and instead force disputes into costly legal proceedings. In 2024, the USPTO agreed to review 70 percent of challenges presented to it; in 2025, that number dropped to 20 percent.
While IPR changes may sound technical, they matter in practice. Limiting access to reviews hands leverage to bad actors who can then more easily use low-quality patents to extract legal settlements. The IPR process does not threaten valid patents on true inventions — it only impacts patents that should not have been issued.
Getting Patent Policy Right
A healthy patent system should not force a choice between patent owners and challengers. Some principles should be widely shared:
- Patents on true inventions should be strong and enforceable. Innovators deserve reliable protection when they develop something genuinely new;
- Invalid or overly broad patents harm competition. Granting monopoly rights beyond the scope of an invention distorts markets and impedes research and development; and
- Efficient review strengthens the system. A timely, expert process for correcting mistakes increases overall confidence in patent rights.
Post-grant review, including IPR, was designed to advance these goals — the current policy changes proposed by the USPTO threaten to undermine them, as well as America’s AI leadership.
Looking Ahead
Policymakers need to chiefly consider outcomes as they weigh the future of patent policy. A system that insulates weak patents from scrutiny risks reviving the very problems Congress sought to fix: higher costs, prolonged disputes, and reduced innovation.
By patenting their own technology and using inventions patented by others, enterprise technology companies are advancing the entire AI ecosystem. This is why it is so important to enforce strong patents on true inventions, while also having an efficient mechanism to weed out invalid patents.
If a bad actor can take a patent that would not hold up in court and use it to extort an AI company or its customers, then the patent system is creating an incentive for trolling, not developing the best technology. We need a patent system that puts the emphasis on invention and innovation, not litigation.