State Farm, USAA and Allstate hold 77% of all insurer AI patents filed since 2014, a 680-patent portfolio built largely on the claim pattern the Federal Circuit invalidated three more times between February 6 and February 24 2026. Each ruling extended the standard the court set in Recentive Analytics, Inc. v. Fox Corp. in April 2025: applying generic machine learning to a known problem, without a concrete technical improvement, is not patent eligible.

77%
Insurer AI Patents Held by 3 Carriers
4
Fed. Cir. Section 101 Invalidations Since Apr. 2025
55
Claims Invalidated in One Feb. 2026 Ruling
31%
Insurer Filings That Are GenAI, Not Traditional ML

Key Takeaways

  • 55 asserted claims across six patents fell in one February 6 ruling, the largest single invalidation of the three, and all three opinions landed inside eighteen days.
  • 77% of insurer AI patents sit with three carriers: State Farm at 326, USAA at 218 and Allstate at 136, out of roughly 883 tracked filings since 2014.
  • "Generic use of AI without other parameters" is abstract, the Federal Circuit held on February 24, which is a description of how most pre-2023 insurer AI claims were drafted.
  • One insurance AI patent has survived a direct application of that standard: Aon Re's claim reciting two independently trained classifiers applied in sequence to aerial imagery.
  • 31% of insurer AI filings were generative AI by October 2025, up from roughly 4% in the early 2020s, and those claims carry more implementation-specific language by necessity.

Three Rulings in Eighteen Days

On February 6 the court affirmed summary judgment in Innovaport LLC v. Target Corp., No. 2024-1545, invalidating all 55 asserted claims across six related patents covering in-store product location. Three days later, in the precedential GoTV Streaming, LLC v. Netflix, Inc., No. 2024-1669, Judge Taranto reversed a jury infringement verdict and invalidated three patents on tailoring television content to a device's capabilities.

The third is the one carriers should read. On February 24, in Rensselaer Polytechnic Institute v. Amazon.com, Inc., No. 2024-1725, a panel led by Judge Dyk affirmed invalidation of US Patent No. 7,177,798, a natural-language patent asserted against Alexa. The court held that "the generic use of AI without other parameters, such as 'improving the mathematical algorithm or making machine learning better,' is abstract."

The priority dates in all three run from 1999 to 2007, well before the current filing wave. Dennis Crouch noted that patents drafted before the 2014 Alice decision "were written in a different era of patent law."

The claim architecture is what carries across. Collect data, apply a known computational technique, output a result is structurally the same as collect insurance data, train a model, output a risk score, which is the pattern that dominates insurer filings from 2018 through 2022. That is the link the Federal Circuit itself drew back to Recentive.

Specificity, Not Subject Matter, Is What Survives

Evident's tracker puts State Farm at 326 patents, USAA at 218 and Allstate at 136, 77% of roughly 883 insurer AI filings since 2014. We covered that concentration in a companion piece. What it did not test is how the claims read against the standard now applied to them.

Two State Farm patents show the exposure. US 11,861,470, a model generation tool for predictive analytics, and US 11,769,213, a method of controlling for undesired factors in machine learning models, both describe building and constraining models in functional, architecture-agnostic terms.

Set against that, Aon Re, Inc. v. Zesty.ai, Inc., No. CV 25-201 (D. Del. July 2025), is the one insurance AI claim tested directly against Recentive and left standing. The court accepted that Recentive bars the mere application of generic machine learning to new data, but found the claims recited "the patent-eligible arrangement of two independently trained classifiers." Claim 1 requires a specific classifier sequence over aerial image pixels in a defined order.

Both claims produce an insurance risk output from a trained model. Aon's says how the model is structured; State Farm's says what it does. That difference is now the whole of it, and it splits the 680-patent base by drafting vintage rather than by carrier.

Claim PatternExample2026 Judicial Exposure
Generic ML applied to a known insurance task (claims triage, severity scoring, pricing)State Farm US 11,861,470; US 11,769,213High. Matches the pattern invalidated in Recentive, GoTV, Innovaport, and RPI v. Amazon
Specific multi-model or classifier arrangement tied to defined dataAon Re's aerial-imagery classifier patent (survived motion to dismiss, D. Del. 2025)Lower. Survived direct application of Recentive
Agentic multi-agent coordination architecturesUSAA's agentic underwriting and claims patentsUntested. No Federal Circuit ruling yet addresses multi-agent claim structures directly
Generative AI content generation and enhancement (aerial imagery clarification, document drafting)USAA's GenAI aerial-imagery enhancement patentsModerate. Depends on whether the claim specifies a concrete generation mechanism or a functional outcome

Generative AI rose to 31% of insurer AI filings by October 2025 from roughly 4% in the early 2020s. Newer architectures give drafters more concrete mechanism to point at, which does not make those filings eligible so much as more likely to read like Aon's claim than State Farm's.

The Moat Was Already Being Priced Lower

A patent portfolio works as a moat by raising the cost of replication, and the threat of an infringement suit is what keeps some carriers building internally rather than buying. Generic, functionally drafted claims no longer deliver that threat reliably, and the vendor market has been acting accordingly.

Sixfold raised a $30 million Series B in January 2026 with Guidewire as a strategic backer, and Munich Re had already integrated Sixfold's underwriting AI into Realytix Zero across intake, pricing and binding. Applied Systems acquired Cytora in September 2025 to fold its risk-processing platform into its own suite. Neither deal was priced as though a claims-triage or risk-scoring suit from a large carrier were a live cost.

The discount is uneven by claim type, which is where it stops being a legal story and becomes a valuation one. Tier 1 generic claims can reasonably be discounted. Multi-agent coordination architectures, the category USAA leads, have no Federal Circuit ruling addressing them at all, so buy decisions there carry unpriced risk in the other direction.

That matters on the balance sheet as well as in procurement. Intangibles account for roughly 90% of S&P 500 market capitalization and more than $62 trillion globally, and acquired AI patents capitalized as identifiable intangibles inherit whatever invalidation risk the claims carry. A seller's warranty that patents are valid and enforceable prices differently after February 2026 than before it, and a portfolio capitalized on durable enforceability is one district court ruling away from an impairment test it was not sized for.

Further Reading

Sources