The billing question generative AI forces is old wine in a new bottle: when technology compresses six hours of work into twenty minutes, what may the lawyer charge? Under ABA Model Rule 1.5, a fee must be reasonable — and ABA Formal Opinion 512 (2024) answers the AI version directly: a lawyer billing by the hour may charge only for time actually spent, so AI efficiency gains belong to the client under an hourly model. That conclusion is not novel; ABA Formal Opinion 93-379 held thirty years earlier that a lawyer who completes work in less time than expected may not bill as if the efficiency had never happened, and that the same research reused for a second client cannot be billed twice. Canada's framework points the same way: FLSC Model Code Rule 3.6-1 requires fees that are fair, reasonable, and disclosed in a timely fashion, with the client's ability to assess reasonableness as a running theme in commentary and assessment (taxation) practice.
So the hourly math is simple, if uncomfortable: bill the twenty minutes of prompting and the ninety minutes of verification and revision — the real lawyer time — not the six hours the task used to take. Padding the entry back to historical norms is a false statement about time worked, which is a discipline problem far beyond a fee dispute. What lawyers may not do is equally clear across both countries; what they should do about the shrinking hour is the strategic question the rest of this article addresses, within our AI in Legal Practice library.
May a firm pass AI tool costs to clients? Yes, within the ordinary expense rules: actual, disclosed, and reasonable. Op. 512 and the general expense analysis in Op. 93-379 combine into a workable framework:
Timekeeping practice needs one further adjustment: describe AI-assisted work accurately in entries. "Draft motion" for eight minutes of prompting plus ninety minutes of verification is defensible if the hours are true; reconstructing entries to disguise how the work was done is not, and fee examiners have become adept at spotting drafting times that are physically impossible for the document produced — in either direction. The clean practice is to record the real activities (criteria design, prompt drafting, output verification, revision) at real durations, which both survives scrutiny and generates the internal data the firm needs to price flat fees intelligently.
Engagement letters should say which model applies. Canadian counsel should remember that accounts are reviewable on assessment, where undisclosed technology charges fare poorly, and that several law societies' fee commentary emphasizes explaining the basis of charges in advance.
The deeper consequence of AI is that it breaks the equation between effort and value that hourly billing assumes. When a first-draft contract takes eight minutes, the hourly model prices the lawyer's most leveraged work at its cheapest — while a flat fee prices the outcome: the client pays for a verified, negotiated agreement, and the firm keeps the efficiency it invested in. That is why AI adoption is accelerating the long-predicted shift toward flat, fixed, and value-based fees, fastest in commoditizable work: routine contracts, uncontested family matters, immigration filings, estate plans, demand-letter programs. Alternative fee arrangements are expressly permitted in both countries provided the total fee remains reasonable (Rule 1.5's factors weigh results, skill, and customary charges — not only hours) and, in Canada, fair and reasonable under Rule 3.6-1 and disclosed up front.
Contingency and statutory-fee practices face their own versions of the question. Contingency fees are unaffected by drafting speed in principle — the fee prices risk and result — but reasonableness review at settlement can still weigh how much lawyer work the percentage actually compensated, and some US courts scrutinizing fee petitions have begun asking whether claimed hours reflect AI-era practice. Statutory and court-awarded fees (class actions, fee-shifting statutes, assessments in Canada) are squarely hours-based, which means the billing-the-time-you-spent rule applies with full force and the supporting time records will be tested against what the technology makes possible.
Two design rules keep flat fees ethical. First, the fee must still be defensible against the reasonableness factors — a flat fee wildly out of proportion to any measure of the work or value can be excessive even if agreed. Second, scope discipline: flat-fee engagements need written scope definitions and change triggers, or the efficiency gains disappear into scope creep. Firms making the transition typically run hybrid books for a year — hourly for bespoke litigation, flat for productized work — and let their own AI-era time data set the prices, which is also the honest way to discover how much AI actually saves (the drafting and review economics are detailed in AI contract drafting and AI document review).
Insurance defense and other audited-billing environments preview where the whole market is heading: billing guidelines from major insurers and corporate legal departments increasingly address AI explicitly — some requiring its use for defined tasks, some capping recoverable hours at AI-era benchmarks for routine work, and nearly all demanding disclosure of tools that touch their data. Firms in these panels are learning the discipline first; everyone else should assume the same expectations arrive with their next institutional client.
Disclosure ties the ethics together. Op. 512 requires candor about AI charges; broader emerging practice — pushed hard by institutional clients whose outside-counsel guidelines increasingly demand it — is to tell clients how the firm uses AI, how it protects their information, and how the economics flow through to fees. The strong position is to volunteer it: "we use AI tools under enterprise confidentiality agreements; on hourly matters you pay only for lawyer time actually spent; on flat-fee matters our efficiency is priced into a lower fixed fee" is a client-development message, not a confession. Corporate clients are already asking; consumer clients soon will, because they are using ChatGPT themselves — often to choose their lawyer, which is why the same firms investing in internal AI should check what AI engines say about them with the free AI Visibility Checker.
The uncomfortable truth for hourly-dependent practices is that competitors will pass AI savings through as lower effective prices, and reasonableness review will increasingly treat AI-era norms as the benchmark. The firms that win the transition are those that move their pricing before the market moves it for them — and that treat billing transparency as part of the same trust posture as verification and confidentiality. For help positioning your firm's growth strategy around AI on every front, book a free strategy call with LexScale.ai.
LexScale.ai helps law firms across Canada and the United States adopt AI for growth — from client-facing intake and content systems to the visibility that puts your firm inside AI answers.
Book a Free Strategy Call →This article is general information, not legal or ethics advice. Professional-conduct rules on AI are evolving and vary by jurisdiction — always verify current requirements with your state bar, law society, or regulator before adopting any AI workflow.
Related Articles
AI Contract Drafting for Lawyers: What Works · AI Document Review & Discovery: From TAR to LLMs · AI Hallucinations in Legal Research: Risks & Fixes · Bar Rules on AI for Lawyers: US & Canada Guidance · Can Lawyers Use ChatGPT? Ethics, Risks & Workflow · Client Confidentiality and AI Tools for Lawyers
Ready to grow your firm with AI?