Time Entry Ethics When AI Did Most of the Work

Time Entry Ethics When AI Did Most of the Work

A contract review that used to take a junior associate four hours now takes 22 minutes with AI assistance. The associate reviews the output, makes corrections, adds judgment, and the final product is arguably better than the four-hour version. So what goes on the time entry? Four hours? Twenty-two minutes? Something in between? And how do you justify whatever number you pick?

This is not a hypothetical. It is the central billing ethics question facing every law firm that has integrated generative AI into its workflow, and the existing rules on reasonable fees were not written with this scenario in mind.

The Reasonable Fee Framework

Model Rule 1.5(a) of the ABA Model Rules of Professional Conduct prohibits lawyers from charging "unreasonable" fees. The rule lists eight factors for determining reasonableness, including the time and labor required, the novelty and difficulty of the questions involved, the skill requisite to perform the legal service properly, and the fee customarily charged in the locality for similar services.

Notice the tension already. "Time and labor required" points toward billing for 22 minutes. "Skill requisite to perform the legal service" and "fee customarily charged" could justify something closer to the historical rate. The rule gives you factors, not a formula, and that ambiguity is where the ethical risk lives.

Several state bars have started weighing in. The Florida Bar issued Ethics Opinion 24-1 in January 2024, which addressed AI use generally but stopped short of giving concrete billing guidance. California's Practical Guidance for the Use of Generative AI in the Practice of Law, published in November 2023, noted that attorneys should consider whether billing for AI-assisted work at traditional rates would result in an unreasonable fee. New York City Bar Association's Formal Opinion 2024-1 was more direct, stating that lawyers may not bill clients for time spent on tasks that AI performed unless the billing reflects the actual time the lawyer spent.

The Value vs. Time Problem

Here is where it gets genuinely interesting. The billable hour has always been a rough proxy for value. A senior partner billing at $1,200 per hour is not really selling time; they are selling judgment, relationships, and pattern recognition built over decades. The hourly rate just happens to be the delivery mechanism.

AI disrupts the proxy. If a tool generates a first draft of a motion to dismiss in three minutes and the attorney spends 45 minutes reviewing, refining, and adding case-specific analysis, the total attorney time is 45 minutes. But the value delivered to the client might be identical to what a pre-AI attorney would have produced in six hours. Billing 45 minutes at the standard rate could actually undervalue the work. Billing six hours is ethically indefensible under any reasonable reading of Rule 1.5.

Some firms are experimenting with hybrid approaches. Susman Godfrey announced in early 2024 that it would pass AI cost savings to clients while maintaining value-based pricing on complex matters. Other firms have quietly begun adding "technology surcharges" to invoices, a practice that raises its own set of questions under Rule 1.5(b), which requires communicating the basis of the fee to the client.

What the Disciplinary Cases Tell Us

We do not yet have a wave of disciplinary actions specifically targeting AI billing practices, but the analogies from existing case law are instructive. In In re Fordham, 423 Mass. 481 (1996), the Massachusetts Supreme Judicial Court found that a fee of $50,000 for a first-offense OUI case was clearly excessive, in part because the attorney billed for time spent learning an area of law he was unfamiliar with. The principle is relevant: you generally cannot bill a client for your own learning curve, and you arguably should not bill them for the time AI saved you from spending.

Missouri v. Jenkins, 491 U.S. 274 (1989), established the lodestar method for calculating reasonable attorney fees in federal fee-shifting cases, multiplying reasonable hours by a reasonable rate. Courts applying the lodestar method in the AI context will inevitably ask whether the hours billed were "reasonable" given the tools available. If opposing counsel can demonstrate that a task should have taken 30 minutes with AI assistance, billing four hours for it becomes a problem even if the attorney actually spent four hours doing it manually.

This cuts both ways, by the way. There is a plausible argument that attorneys who refuse to adopt efficiency-enhancing technology and then bill clients for the extra time are themselves charging unreasonable fees. The ABA's Formal Opinion 477R (2017) already established a duty of technological competence under Model Rule 1.1, Comment 8. It is not a stretch to connect competence obligations to billing practices.

Practical Guidance for Right Now

Given the regulatory ambiguity, firms need internal policies before the bar associations finish deliberating. A few concrete recommendations:

• Track AI-assisted time separately. Your timekeeping system should distinguish between attorney work time and AI processing time. If you cannot reconstruct how a bill was generated, you cannot defend it.
• Disclose AI use in engagement letters. Several state bars, including California and New York, are moving toward requiring this. Getting ahead of the requirement protects you and builds client trust.
• Bill for attorney judgment, not AI output. The defensible position is billing for the time the attorney actually spent reviewing, analyzing, and refining AI-generated work. The AI output itself is a tool cost, like Westlaw or document management software.
• Consider value-based fee arrangements. Fixed fees, capped fees, and success fees sidestep the hourly billing problem entirely. Clients increasingly prefer them, and they align better with AI-augmented workflows.
• Audit your invoices for indefensible entries. "Research and drafting: 6.0 hours" on a routine contract review is going to draw scrutiny from sophisticated clients and, eventually, from disciplinary committees. Internal review before invoices go out is essential.

The Corporate Client Perspective

In-house legal departments are not waiting for ethics opinions. A 2024 survey by the Association of Corporate Counsel found that 57% of chief legal officers expected outside counsel to pass along AI-driven cost savings. Major legal operations teams at companies like Microsoft, JPMorgan Chase, and Johnson & Johnson have begun requiring outside counsel to disclose AI usage in billing guidelines. Some are using their own AI tools to audit invoices for reasonableness, comparing billed hours against benchmarks for AI-assisted work.

If your client's billing review software flags your invoice before you flag it yourself, you have already lost credibility. And credibility, once lost with a sophisticated in-house team, does not come back easily.

The Regulatory Trajectory

The ABA's Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512 in July 2024, addressing various aspects of generative AI in legal practice. While it touched on confidentiality and competence obligations more than billing specifics, the opinion reinforced that existing ethical rules apply fully to AI-assisted work. More targeted billing guidance is expected in 2025, and several state bars have active working groups on the question.

The direction is clear even if the specific rules are still forming. Transparency, actual-time billing, and client communication will be the baseline expectations. Firms that build these practices now will not need to retrofit when formal opinions arrive.

How FirmAdapt Addresses This

FirmAdapt's platform includes granular audit logging that tracks every AI interaction, the prompts submitted, the outputs generated, and the attorney review time applied to each task. This creates a defensible record for billing purposes, allowing firms to demonstrate exactly how much attorney time went into a given work product and how AI tools were used in the process.

The platform also supports configurable billing policies that can enforce firm-specific rules about how AI-assisted work is categorized on invoices. If your policy is to bill only for attorney review time on AI-drafted documents, FirmAdapt can flag time entries that deviate from that standard before they reach the client. The compliance architecture is designed to make the right billing practice the default, not an afterthought.