The AI Patent Litigation War Hasn’t Started—But It’s Coming Fast

In certain industries, especially in software and technology, there is often a delay between products appearing on the market and patent litigation being filed on those products. We’re likely in the midst of a similar delay for artificial intelligence-related technology.

The number of patents filed under CPC Subclass G06N, which covers “computing arrangements based on specific computational models” and includes “computing arrangements based on neural networks” and “machine learning,” have soared since 2007. More than 32,000 patents with this classification were granted in 2025, compared to 426 in 2008, according to a Derwent search. However, only 53 district court lawsuits filed in 2025 where that specific classification was asserted, according to DocketNavigator.

If AI follows the pattern of smartphones, where patent litigation drastically increased in the several years after commercial launch, we likely will see patent litigation in this space increase significantly over the next several years.

These cases span diverse technologies, with patents directed to detecting vaping and bullying, predicting human interaction with vehicles, and echocardiographic image analysis. According to DocketNavigator, where there has been an outcome in the district court, the vast majority of cases have either settled or have been voluntarily dismissed, or the accused infringer prevailed on establishing invalidity.

The results at the Patent Trial and Appeal Board have been more mixed, with patent owners at least partially prevailing a majority of the time. This suggests that Section 101 invalidity presents a larger issue for these patents than Sections 102 or 103.

With that in mind, companies in this industry should consider spending time preparing for eventual litigation.

One step AI companies should consider is evaluating what intellectual property protection they have and whether it provides adequate coverage and was documented sufficiently. They will want to evaluate both the scope of trade secret protection (especially for confidential algorithms, training data, and modeling) and patent protection.

Companies should evaluate whether their trade secrets are adequately defined and protected. In Applied Predictive Technologies, Inc. v. MarketDial, Inc., the US Court of Appeals for the Federal Circuit affirmed summary judgment against the plaintiff for not sufficiently identifying and defining the alleged trade secrets.

They also need to determine that trade secrets have been protected in all necessary agreements with employees, vendors, independent contractors, and any joint developers, and have enough security around the trade secrets material.

For patents, companies, likely in-house counsel working with outside counsel, should evaluate the scope of their portfolio, including how many patents they hold and the collective scope of patent coverage. Are there gaps in patent coverage for technology that should be protected? Companies should ensure that ownership and assignment records are in place and are legally sufficient.

For particular patent families—perhaps patents covering key technology or considered commercially relevant—companies should evaluate claim language for clarity and scope. Claims also should be evaluated for patent eligibility under prevailing law. Last year, the Federal Circuit held in Recentive Analytics Inc. v. Fox Corp. that applying generic machine-learning techniques in a particular environment alone doesn’t make a claim patent eligible.

Outside of IP, companies should consider spending some time aligning documentation and processes for potential litigation. Evaluate what documentation and materials exist that show development of their products, software, and technology. This is an area that may be especially difficult to evaluate for AI-focused companies. Has the team been documenting how certain problems were solved and if iterations of other solutions were tested? If technology is developed jointly, where are materials stored and who has access to them?

Companies should evaluate how their technology is being described and marketed to ensure descriptions are accurate and aligned with the company’s IP strategy. They should make sure there are adequate processes for when employees depart, both for ongoing protection of trade secret information and potential cooperation in the event of future litigation.

Patent litigation in this field hasn’t yet begun in earnest, but it almost certainly will be coming in the next few years. Companies can spend some time now getting ready.

This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.

Author Information

Alexis Cohen is an intellectual property partner at Sidley Austin in Washington.

Interested in writing? Review our author guidelines, and submit pitches to Insights@bloombergindustry.com.