The “Verification-Value Paradox”:  No Evidence That Counsel’s Use of AI Should Have Decreased Billing

Much has been written about the impact of A.I. on attorney’s hourly billing.  That issue was addressed in the lengthy opinion in Flux Drive, Inc. v. Rong, __ F.4^th __, 2026 WL 3266333 (10^th Cir. Oct. 30, 2026).  The court rejected an argument that, because A.I. was used, fewer hours should have been billed.

THE VERIFICATION-VALUE PARADOX

It is now axiomatic that AI output must be verified.  The cost of verification of the results of an AI search may be significant.  Lee Vickery has written “When Efficiency Meets the Duty to Verify: Reflections on the Verification-Value Paradox” (Jan. 1, 2026)(paywall for Law.com). He explained:

The Verification-Value Paradox states that increases in efficiency from AI use “will be met by a correspondingly greater imperative to manually verify” the outputs. The result is that the net value of AI in many legal contexts may be negligible once verification is honestly accounted for. For low-stakes tasks, verification costs are light. For core legal work, verification costs are heavy. That’s the tension. [emphasis added].

In his blog, The Verification-Value Paradox (Jan. 12, 2026), Doug Austin traces the concept to Joshua Yuvaraj’s article, The Verification-Value Paradox: A Normative Critique of Gen AI Use in Legal Practice (Oct. 23, 2025). Doug wrote:  “It is often claimed that machine learning-based generative AI products will drastically streamline and reduce the cost of legal practice….  Cases in Australia and elsewhere in which lawyers have been reprimanded for submitting inaccurate AI-generated content to courts suggest this paradigm must be revisited.”

Mr. Yuvaraj wrote that the verification-value paradox “suggests increases in efficiency from AI use in legal practice will be met by a correspondingly greater imperative to manually verify any outputs of that use, rendering the net value of AI use often negligible to lawyers….  ”  The article suggests that the claimed value of GenAI (as opposed to predictive AI) may be negligible in the practice of law.  To state the obvious, “[v]erification is costly” and “it is essential to the practice of law….”

The paradox is grounded in the following logic: the net value of an AI model in legal practice can only be assessed once the efficiency gain (savings on time, salary costs, firm resource costs, etc) is offset by the corresponding verification cost (cost to manually verify AI outputs for accuracy, completeness, relevance, etc)…. The verification-value paradox suggests that the more that needs to be manually verified, the less the efficiency gain….  In a text-dominated profession, the greater the AI use, the greater the cost of manual verification.

The article states: “This paradox is not novel in the sense of presenting new knowledge: many scholars, practitioners and judges have articulated similar problems with AI and emphasised the importance of verifying AI outputs.”  The author summarized: “The verification-value paradox suggests the net value of AI to legal practice is grossly overestimated, due to an underestimation of the verification cost. A proper understanding the costly and essential nature of verification leads to the conclusion that AI’s net value will often be negligible in legal practice: that is, in most cases, the value added will not be sufficient to justify the corresponding verification cost.”  One caveat stated by the author is that the paradox “is fundamentally theoretical, even though anecdotal evidence seems to bear it out. The paradox is therefore a hypothesis for further examination.”

Mr. Yuvaraj wrote: “Studies have repeatedly found ‘hallucinations’ even in machine learning tools built for the legal context.” [citation omitted].  For more on that topic, please visit:

• “Hallucinations” by West’s CoCounsel? (Apr. 7, 2026);
• “Hallucinations” by West and Lexis AI?  A Cautionary Study and Cautions About the Study (Apr. 11, 2026)(suggesting that this 2024 study be read with caution in 2026).

With that background, I turn to the facts of Flux Drive.

THE VERIFICATION-VALUE PARADOX APPLIED: 

PLAINTIFF ASSERTED THAT DEFENDANT’S USE OF A.I. SHOULD REDUCE HOURS

Flux Drive involved three consolidated appeals, and a lengthy analysis that I make no effort to summarize.

In pertinent part, the decision addresses a challenge to the reasonableness of awarded attorneys’ fees in the amount of $114,663.11.  The case turned on allegations that Defendants “conspired to steal [Plaintiff’s] trade secrets and transfer them to a company in China that manufactures unauthorized knockoff products.”

The Tenth Circuit affirmed the award, under Colorado law, of the full amount requested by certain Defendants.  Applying the lodestar method, the court noted that the reviewing court must ensure that the winning attorneys exercised billing judgment “by winnowing out unreasonable hours.”

On appeal, Plaintiffs challenge the reasonableness of the fee award on multiple grounds. They contend that defense counsel: … (3) should have billed fewer hours due to their reliance on artificial intelligence (‘AI’) and other online research tools. [emphasis added].

PLAINTIFF’S ASSERTION WAS REJECTED

BASED IN PART ON VERIFICATION COSTS

The Tenth Circuit wrote:

On their third argument, Plaintiffs identify no evidence that counsel’s use of AI resulted in excessive billing. It is possible that a law firm’s use of AI and other online research tools may “reduce[ ] the number of hours required” for research and thereby lower the lodestar amount…. As the magistrate judge explained, however, ethical rules require counsel to verify the cases cited and to ensure that all arguments are grounded in legal authority…. Thus, even assuming that AI and other research tools might make attorneys more efficient in the long run—and, in some circumstances, might thereby reduce the hours necessary to litigate a case—Plaintiffs have not shown that defense counsel’s use of AI rendered the hours that they actually expended unreasonable.

Id. at *15 (emphasis added).  The fee award for full fees was affirmed.  The Verification-Value Paradox appears to have “legs.”