Clients want speed and solid work. They don’t want mystery “AI fees” buried on an invoice. So, can you bill for AI in 2025? Yes—when you charge for your legal skill and oversight, not for a robot button press.

Here’s the simple version: follow the ethics rules, be clear about costs, and keep your supervision tight. We’ll connect the dots between reasonable fees (Rule 1.5), confidentiality (1.6), competence (1.1), and vendor supervision (5.3) so you can recover AI costs without drama.

What we’ll cover:

• What you can bill for vs. what you can’t (your judgment and review vs. “AI time”)
• Cost recovery: passing through AI expenses at actual cost vs. treating subscriptions as overhead
• Disclosure and client consent for AI in legal services—and when it’s required
• Billing models that fit AI-enabled work (hourly narratives, flat fees, AFAs, value-based pricing)
• Engagement letter language, documentation, and audit trails to avoid markups on AI disbursements
• Data confidentiality and vendor management fundamentals
• How LegalSoul helps operationalize ethical AI billing, from usage tracking to clear invoices

Quick answer: Can lawyers bill for AI use in 2025?

Short answer: yes. Bill for your legal analysis, strategy, and quality checks—not “AI time.” Recent guidance (Florida Bar Ethics Op. 24-1; North Carolina 2023 FEO 3) says fees must be reasonable, lawyers must stay competent with the tech, protect client info, and supervise nonlawyer tools and vendors.

What works in practice: tell clients when AI use is material, capture per-matter tool costs, explain your oversight in time entries, and pick fee structures that reward efficiency. Example: AI outlines a brief; you verify citations, refine arguments, and tailor to the forum—that’s billable. If there’s a per-document charge tied to the matter and your letter allows it, you can pass that cost through at actual cost. Fair to clients and fair to your team.

Quick Takeaways

• Bill for legal value. Charge for judgment, editing, and review—never for “AI time.” Keep to Model Rules 1.1, 1.5, 1.6, and 5.3, and disclose when AI use matters.
• Treat costs the right way. Pass through matter-specific AI fees at actual cost with consent; subscriptions usually live in overhead. No secret markups.
• Choose pricing that fits. Clear hourly narratives, flat fees for repeat work, and portfolios or value-based options beat vague “AI surcharges.”
• Show your work. Keep usage logs, audit trails, and review notes. Supervise outputs and vendors. LegalSoul can help with tracking, disclosures, and clean invoices.

Why this matters now: 2025 market dynamics and client expectations

In-house teams are asking for speed and accuracy without surprise add-ons. Over the last year, many have pushed back on “AI surcharges,” and they’re treating most tech platforms as firm overhead unless a specific matter creates a direct tool expense.

Buyers want predictability and clear value. One Am Law pricing lead told the trades they dropped the AI line item, moved to fixed-fee bundles for research-heavy projects, and both realization and client satisfaction went up. RFPs often ask how you use tech but say “no markups.” If you’ve been piloting AI, 2025 is the year to align your message and your pricing and make your controls obvious.

Ethics framework for AI billing (Model Rules 1.1, 1.5, 1.6, 5.3)

Everything flows from the rules. Rule 1.1 says be competent with the tech—including limits like hallucinations and data exposure. Rule 1.5 requires reasonable fees; ABA Formal Op. 93-379 warns against undisclosed markups on disbursements and treats general tech as overhead.

Rule 1.6 covers confidentiality—watch vendor terms, retention, and access. Rule 5.3 requires supervising nonlawyer assistance, which includes AI tools. Florida 24-1 and North Carolina 2023 FEO 3 both allow AI with proper oversight, fair fees, and informed consent for expenses. Tie your internal review standards to your fee model and document what you check. Your invoices will stand up better in audits.

What you can bill for vs. what you can’t

Bill for the parts that need a lawyer’s brain: analysis, strategy, tailoring to jurisdiction, citation checks, drafting and edits, and quality control. Don’t bill “AI time,” and don’t charge old manual hours when AI genuinely saved time.

Try this test: does a competent lawyer need to exercise judgment here? If yes, it’s billable. If AI just counted words or reformatted a table, that’s not legal work. If a client bans AI, respect it and don’t sneak it in. And if an NDA draft takes two hours with AI instead of five, bill the two and consider flat-fee options for similar work going forward.

Cost recovery: Pass-through expenses vs. overhead

Passing costs through depends on whether they’re tied to the matter and properly disclosed. Under ABA Op. 93-379 and similar state opinions, you can pass through disbursements at actual cost with informed consent. No hidden markups. Subscriptions and platform licenses are usually overhead unless your engagement letter says otherwise.

Examples: your AI tool charges $8 per long-document analysis and you scan 10 contracts in diligence. If your letter authorizes pass-through at cost, list “AI document analysis (10) @ $8 = $80.” No markup. But a $1,000 monthly AI drafting suite? Unless agreed otherwise, that’s overhead—bake it into your rates or flat fees. Pro tip: create a billing code for AI pass-throughs and require a short usage note tied to the task.

Disclosure and informed consent: When and how to tell clients

Disclose when AI use is material or when you’ll incur a client expense. The exact threshold varies, but it’s smart to set expectations early: how you use AI, your safeguards, and any likely per-matter costs you’ll pass through at actual cost.

Add a clear AI section in your engagement letter: potential uses (research, drafting, review), attorney supervision, confidentiality measures, and permission to pass through specific usage fees at cost. If you switch tools or data routing changes, send an update. On invoices, avoid vague “AI fee.” Use plain labels like “Per-document analysis fee at actual cost.” Clients and procurement teams will appreciate it.

Billing models that align with AI-driven efficiency

Price the value, not the minutes. Four models tend to fit:

(1) Hourly with good narratives—explain the decisions you made and what you verified. (2) Flat fees for repeatable tasks—predictable for clients, better margins for you as the work gets faster. (3) Portfolios or subscriptions for ongoing workflows like contracts or compliance. (4) Value or success fees where allowed.

Many buyers prefer fixed or outcome-based pricing for AI-heavy work and say no to separate AI charges. A simple approach: tiered flat fees based on complexity, with a promise of no surprise tech expenses. You can also package fast-turn services—like a 48-hour “contract risk snapshot”—and price for speed-to-insight.

Engagement letter language and policy templates

Clean language avoids haggling later. Consider including:

(1) Scope and AI use: “Firm may use vetted AI tools for research, drafting, and document analysis under attorney supervision.” (2) Confidentiality: “No training on client data; strict access controls; retention per policy.” (3) Expenses: “Client approves pass-through of matter-specific AI usage fees at actual cost; general subscriptions are Firm overhead unless we agree otherwise.” (4) Client preferences: opt-out options or limits. (5) Backup: “Firm can provide reasonable support for pass-through expenses on request.”

Match your internal policy to this: approved tools, review standards, data handling, and billing narratives that highlight analysis and oversight—not “ran AI.” Add turnaround SLAs for flat-fee or portfolio work. Clients love clarity.

Documentation and audit trails to support reasonableness

When questions come up, your records should make the answers easy. Keep matter-level logs of what AI was used for, and how a lawyer reviewed and fixed or accepted outputs. If you pass through a fee, keep the platform’s usage proof and tie it to a specific task.

Time entries should show the professional work: “Reviewed AI research; validated authorities; refined argument; weighed adverse cases.” Avoid entries that sound like you were just pushing buttons. Quick QA notes—what you checked, what changed, why—can save your invoice during a rate review. Even on flat fees, keep the trail for competence and supervision.

Data security and vendor risk management

Expect clients to ask where data goes and who can see it. Vet vendors on key points: no training on your data, encryption in transit and at rest, retention windows, data location, and whether you can route to private or enterprise models. Put deletion rights in the contract.

Inside the firm, control inputs, apply role-based access, and default to redaction. Test models with planted errors to confirm your review process actually catches problems. For especially sensitive work, consider an isolated environment. Bars in Florida and California have stressed confidentiality and supervision—treat vendor reviews as ongoing, not a one-off.

Jurisdictional highlights and evolving ethics opinions

Across states, the main ideas match: AI is allowed with competence and supervision; fees must be reasonable; disclose material use and get consent for expenses; protect confidentiality. Florida Bar Ethics Op. 24-1 (2024) and North Carolina 2023 FEO 3 both emphasize these points. California’s guidance echoes fair fees, privacy, and vendor diligence.

Differences exist. Some places lean harder on disclosure for substantive drafting; others focus on confidentiality and vendor terms. Keep a quick tracker for your main jurisdictions and build your policy to the highest standard so you don’t need to rethink it matter by matter.

What in-house counsel want to see on invoices

They want to see the legal value and the checks you performed, not a mysterious AI line. Strong narratives look like: “Analyzed AI memo; confirmed citations; applied facts to holdings; drafted strategy; partner review for risk.” If you pass through a fee, label it clearly—“AI document analysis fee at actual cost”—and keep the receipt.

Avoid entries that read like tool operation. For portfolios or flat fees, describe milestones and outcomes. A one-line note like “Turnaround accelerated via approved AI under attorney supervision” signals competence and cost care. Track pushback on entries referencing AI, then adjust your language.

Common pitfalls to avoid

• Hidden markups on AI expenses. Pass through at actual cost with consent or don’t pass them through.
• Charging subscriptions to clients. Unless agreed otherwise, that’s overhead. Adjust rates or flat fees instead.
• Skipping lawyer review. Courts have rejected filings with unverified AI citations. Always verify.
• Vague engagement letters. If you plan to recover per-matter AI fees, say so upfront.
• Weak billing narratives. “Used AI” isn’t persuasive. Explain the legal work.
• Data leaks. Disable training on client data, limit retention, and control inputs.
• Inconsistent internal rules. Set one policy for pass-throughs and follow it.

Quiet metric worth tracking: how often attorney review corrected or improved AI output. That’s your value, and it’s hard to argue with.

Implementation roadmap for your firm

• Map your workflows. Sort tasks by risk and repeatability. Flag where AI helps without cutting quality.
• Pick pricing that fits. Flat fees or portfolios for repeat work; hourly for bespoke strategy. Test value-based where allowed.
• Refresh engagement letters. Add AI disclosures, confidentiality terms, and pass-through authorization at actual cost.
• Set billing codes. One for internal AI use (non-billable), one for pass-through expenses with a required usage note.
• Train and certify. Short courses on supervision, data handling, and writing better narratives. Require sign-off before access.
• Pilot and measure. Start small, track cycle time, realization, and client feedback. Adjust and expand.
• Audit regularly. Quarterly reviews of logs, invoices, vendor list, and retention settings.

One more tip: publish turnaround targets for AI-enabled matters and match pricing to those speeds. Clients will pay for predictability.

How LegalSoul helps you bill ethically and confidently

LegalSoul makes the mechanics simple. It tracks matter-level AI usage so you can pass through fees at actual cost with clean proof. Prompts, outputs, and attorney review notes get logged, showing your supervision without extra hassle.

You also get engagement letter clauses for disclosure, consent, and expense recovery that line up with current opinions. Time and billing integrations turn your work into narratives that highlight legal analysis and oversight. Pricing dashboards show cycle-time gains and realization so you can choose hourly, flat, or portfolio models wisely. Fewer write-offs. Faster approvals. Happier clients.

FAQs

Can I recover a portion of my AI subscription?

Usually no. Subscriptions are overhead unless your engagement letter clearly says otherwise. You can pass through per-matter usage fees at actual cost with consent.

Do I need client consent for every AI use?

Not for every click. Disclose material use and get consent before any client expense. If data is sensitive, talk to the client first.

How do I bill if AI drafts 80% and I refine 20%?

Bill for what you did: review, verification, tailoring, strategy. Don’t bill “AI time.” If you see this pattern a lot, consider a flat-fee approach.

What if a client bans AI?

Follow the ban, document it, and adjust timelines and pricing. Offer a non-AI workflow so expectations are clear.

Key takeaways and next steps

• Charge for legal work, not tool operation. Your review and judgment are the value.
• Get costs right. Pass through matter-specific fees at actual cost with consent; keep subscriptions in overhead unless agreed.
• Be clear and keep records. Disclose material AI use, update letters, and maintain logs and review notes.
• Pick smart pricing. Hourly with strong narratives for bespoke matters; flat or portfolio deals where work repeats.
• Protect data and supervise vendors. Keep confidentiality front and center.

Next steps: update your engagement templates, add billing codes for pass-throughs, run a pilot with AI-enabled work under a flat or portfolio model, and track client feedback and realization. With clear rules and transparent billing, you’ll capture the upside and earn trust.

Conclusion

You can bill for AI-assisted work in 2025—just keep fees fair, disclose what matters, and make sure a lawyer is supervising and protecting client data. Charge for judgment and review, not “AI time.” Pass through per-matter fees at actual cost (with consent) and treat subscriptions as overhead unless the client agrees otherwise. Favor pricing that rewards efficiency and keep good records to back it up. Want a fast way to put this into practice? Use LegalSoul to track usage, generate clean narratives, and plug in ready-made clauses clients will approve. Start a small pilot with LegalSoul and see how it feels.