Allstate's newly granted US Patent 12,711,555 B2 runs a flagged policy through a second premium model, one built with different modeling techniques than the filed rating plan, and when that alternative price and the standard price diverge past a set threshold, the system rejects the policy and fires the notification itself. The U.S. Patent and Trademark Office issued the grant on August 18, 2026, with 17 claims covering the mechanism (Google Patents, USPTO, August 2026).

That is not a claims chatbot or a customer-service assistant, the categories that dominate most insurer AI patent filings. It is a pricing and risk-selection engine, and it places an automated accept-reject decision squarely inside actuarial rate adequacy, the question of whether the premium a policy will pay covers the cost of the risk it represents. For pricing and underwriting teams, the question the grant raises is not whether Allstate can build this; the patent proves it already has. The question is what happens to a book of business when a machine, not an underwriter, draws the adequacy line, and how a carrier documents that decision when a regulator or a rejected applicant asks why.

Inside Claim 1: Three Modules and a Threshold

US 12,711,555 B2, titled "System and Network for Tiered Optimization," names three Allstate engineers, David MacInnis, Jennifer Jabben, and Teresa J. Dalenta, all based in the Chicago suburbs, and assigns the patent to Allstate Insurance Company of Northbrook, Illinois (Google Patents, USPTO, August 2026). Claim 1 describes a system built from three communicating modules. An inspection module retrieves inspection information tied to a physical inspection of the insured asset, a plurality of "rejection conditions," and runs an inspection optimization algorithm that decides whether to waive mitigation of any condition it finds. When the algorithm determines that a condition requires mitigation to keep the policy in force, a second component, the unrestrained premium module, generates what the patent calls an "unrestrained rating plan premium," built from a plurality of premium components "using first modeling techniques different from second modeling techniques used to generate the policy premium" (US 12,711,555 B2, Claim 1). A third component, the underwriting module, takes that unrestrained premium and the standard policy premium, computes a score from the two, compares the score to a threshold, and, in the claim's own language, is configured to, "responsive to determining that the score does not meet the score threshold, automatically reject the policy and transmit a rejection notification to a computing device associated with the policy" (US 12,711,555 B2, Claim 1).

Two comparisons are doing the work. The inspection module compares a "lifetime premium with loss" against a "lifetime premium after mitigation," and if the loss-scenario premium is cheaper than what mitigation would cost, the system waives the mitigation requirement outright and marks the policy accepted with conditions; if mitigation is cheaper, it stays required (US 12,711,555 B2, USPTO, August 2026). The underwriting module then runs the second comparison, checking the standard rating-plan premium against the unrestrained premium the second module built independently. Dependent claim 3 specifies that the underwriting module sets its score threshold "based on a ratio between the policy premium and the unrestrained rating plan premium and is based on monitoring of one or more underwriting factors within a geographical region," meaning the bar for rejection can move by market (Claim 3). Claim 2 confirms the premium components behind both models include "at least one multiplicative factor and at least one additive factor," the standard building blocks of a rating algorithm (Claim 2).

A Three-Tier Rejection Ladder, Not a Single Cutoff

The patent's description walks through a more granular process than claim 1's single threshold suggests. The underwriting module first sorts every rejection condition it finds into one of three priority-based profiles, ranked by the risk, probability, or severity the condition represents, before the premium comparison even runs.

Priority tier Trigger System action
Priority 1 (high) At least one priority-1 rejection condition present Immediate rejection; rejection letter lists all applicable conditions, no premium comparison run
Indicator-based (medium) No priority-1 condition present Underwriting module requests the unrestrained premium, calculates the score, and compares it to the threshold; a score that clears the threshold accepts the policy even with a low-priority condition present
Priority 3 (low) Score fails to meet the threshold System checks for priority-3 conditions; if present, rejects and lists them; if absent, rejects anyway based on the medium-priority condition alone

The design point is that a bad-enough condition never reaches the pricing model at all, it is rejected on sight, while a borderline condition gets one chance to clear the price-based test before the system falls back to rejecting on the condition itself. A parallel wind pool module runs the identical unrestrained-premium comparison to decide whether wind coverage must be mandated, excluded, or left alone on a given policy, with its own threshold that the specification says "may vary by coastal bands" (US 12,711,555 B2, USPTO, August 2026).

Nine Years From First Filing to This Grant

The idea is not new at Allstate; the claim language just widened. Application 18/585,776, filed February 23, 2024 and granted as 12,711,555 B2, is a continuation of application 15/245,343, filed August 24, 2016 and granted in March 2024 as US 11,928,736 B2, also titled "System and Network for Tiered Optimization" (Google Patents, USPTO). The 2024 parent patent's abstract and claims describe the mechanism specifically for a "new homeowner insurance policy." The August 2026 continuation drops that limitation from its independent claims, describing the same architecture generically as applying to "a policy," a broadening that extends the same accept-reject machinery to other lines without Allstate having to file, and prosecute, an entirely new application. Examiner Hai Tran handled prosecution; the parent's own file history cites a Canadian counterpart, CA 3,034,867, and an international application, WO 2018/039321, filed the same month as the original US application, indicating Allstate sought protection for the mechanism outside the United States from the start.

Allstate is not filing AI patents in a vacuum. Patent-tracking firm Evident found that State Farm, USAA, and Allstate together account for 77% of all AI-related patents filed by insurers since 2014, State Farm with 326, USAA with 218, and Allstate with 136, and that P&C insurers as a class hold 89% of every AI patent the sector has filed in that span (Insurance Journal, December 2025, citing Evident). "Patents offer a rare window into where insurers are placing their biggest bets on AI," Evident chief executive Alexandra Mousavizadeh told the trade press (Insurance Journal, December 2025). Most of that broader portfolio still skews toward claims and customer service; actuary.info's analysis of the Evident concentration data found generative AI filings surged from 4% of insurer AI patents in 2014 to 31% by October 2025, nearly all of it aimed at claims handling and service chat, not pricing. A patent that automates the accept-reject line itself, rather than a document-summarization or claims-triage task, is a narrower and more consequential category inside that count.

What "Unrestrained" Means to an Actuary

The mechanism is a rate-adequacy test automated at the point of underwriting, not a marketing or customer-acquisition tool. A standard rating plan is the filed set of factors, relativities, and rate caps a regulator has approved; it is, by construction, restrained by whatever the state's prior-approval or file-and-use process allows. An "unrestrained" premium, built with a distinct modeling technique, is not bound by those same filed constraints, so it can register a fuller cost indication for a given risk than the filed plan is permitted to charge. Comparing the two and rejecting outright when the gap is too wide is functionally a real-time solvency and rate-adequacy check: if the risk the model actually prices is too far above what the filed plan can legally charge for it, the system declines to write the business rather than accept an inadequate premium.

That is a different exercise from the "price optimization" regulators moved against a decade earlier, when several states restricted using a policyholder's price elasticity or willingness to pay, factors unrelated to the cost of risk, to set premium levels above a cost-justified rate. The NAIC's Casualty Actuarial and Statistical Task Force finalized its white paper on price optimization in November 2015 after extended public comment, warning regulators that rating plans built around demand elasticity rather than cost could violate unfair-discrimination standards (NAIC Casualty Actuarial and Statistical Task Force, November 2015). Allstate's patent describes the opposite direction, testing whether the filed premium is too low relative to an unconstrained cost estimate, not raising it above cost to capture willingness to pay. The distinction matters for how a regulator is likely to read the filing, but it does not remove the governance question the patent creates: an automated system is now deciding, without a human underwriter in the loop under claim 1, whose risk the carrier is willing to accept at the price it is legally allowed to charge.

Governance Exposure: An Automated Reject Meets the AI Bulletin

The NAIC's Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted December 4, 2023, names underwriting explicitly as one of the insurance lifecycle stages its AI Systems Program requirements cover, alongside pricing, claims, and marketing (NAIC, December 2023). The bulletin defines an "Adverse Consumer Outcome" as any decision "subject to insurance regulatory standards enforced by the Department that adversely impacts the consumer in a manner that violates those standards," and it ties the Department's oversight authority to, among other statutes, the Property and Casualty Model Rating Law's requirement that rates not be "excessive, inadequate, or unfairly discriminatory" (NAIC Model Bulletin, December 2023). A system whose entire claimed function is producing automated rejections is generating exactly the kind of consumer-impacting decision the bulletin's documentation expectations are built around: insurers are expected to be able to reconstruct why a given policy was declined, what data and thresholds drove it, and whether the outcome was tested for disparate impact.

Claim 1 of Allstate's patent does not require a human review step before the underwriting module transmits its rejection notification. The wind pool module and inspection module both offer conditional paths, accept with conditions, waive mitigation, that soften a straight reject, but none of the independent claims require a person to see the file before the system acts. That leaves the documentation burden the NAIC bulletin describes resting entirely on records the system itself must generate and preserve: the unrestrained premium's inputs, the threshold in effect for that policy's geography, and the rejection condition, if any, that triggered the outcome. A carrier that cannot reconstruct those inputs on request from a market-conduct examiner is exposed regardless of whether the underlying rejection was actuarially sound.

Build Versus Buy for Everyone Who Isn't Allstate

A granted patent on an automated accept-reject engine changes the calculus for carriers that lean on third-party rating and underwriting vendors rather than building pricing infrastructure in-house. Services firms with substantial insurance AI patent portfolios of their own, EXL has filed ten insurance-specific AI patents covering document processing and underwriting infrastructure, for instance, now have a concrete claim set to design around if they want to sell a comparable tiered-optimization capability to Allstate's competitors. A mid-market or regional carrier evaluating whether to build its own unrestrained-premium comparison engine, license one from a vendor, or keep a human underwriter making the final call has to weigh that decision against the risk that the cleanest version of the mechanism, comparing a filed premium to an unconstrained one and auto-rejecting on the gap, is now owned intellectual property. actuary.info's look at AIG's underwriting patent portfolio found the same dynamic playing out on the claims-triage side: carriers with large in-house AI patent estates gain a structural edge over the rest of the market's technology roadmap, not just a marketing claim.

The timing lands as Allstate is actively pushing the book this mechanism would govern. Homeowners new premiums written rose 8.1% to $4.8 billion in the second quarter of 2026, and the line's combined ratio improved to 94.6 from an unprofitable 102 a year earlier, while property-liability underwriting income overall reached $2.0 billion, up 56.7% (Insurance Journal, August 2026, citing Allstate's second-quarter results). Growing a book while a patented engine is simultaneously authorized to auto-decline any policy whose unrestrained price falls too far below the filed rate is not a contradiction, it is the adequacy discipline that makes profitable growth durable rather than a one-quarter reserve release. But it puts more weight on the audit trail behind every algorithmic no than a carrier growing new business 8.1% a quarter can afford to leave undocumented.

Further Reading

Sources