AI's Impact on Lawyers, Accountants, and Tax Advisors: What Changes, What Stays

In 2023, a lawyer filed a court brief written with the help of ChatGPT — and every case it cited was an AI fabrication (non-existent cases). The lawyer was sanctioned, and the episode spread global wariness about "law and AI." Yet ironically, in the years since, AI adoption among professionals exploded. By one survey, over 90% of lawyers already use some AI tool in their daily work. Feared, yet used anyway — few professions are tested on their relationship with AI as much as these are.

Here's the bottom line. AI fiercely streamlines the "research, draft, and record" work of lawyers, accountants, and tax advisors, while the professional core — judging, taking responsibility, and signing — stays with humans. Whether an audit opinion or legal advice, it carries legal liability and professional skepticism, and that cannot be delegated to AI. As the next entry in our "AI impact by industry" series after #068 (trading companies), #094 (marketing), and #097 (consulting), this article surveys what's happening in the professions: the impact by role, the biggest pitfall, the unchanging core, the junior crisis, and advice by role. Note that frameworks of regulation and liability differ by country, and all figures are vendor/survey-published values.

* The adoption, time-savings, accuracy, and employment figures in this article are citations of vendor/survey-published values (as of 2026, many US-based) and include best-case or specific-segment numbers. Systems, liability, and regulation differ greatly by country; in practice, follow your jurisdiction's laws and ethics rules.

1. The state of play, in numbers

"The professions are changing with AI" is no longer a forecast — it's a reality in progress. First, let's gauge the temperature with the reported numbers (all published, condition-dependent).

• Adoption: by one survey, over 90% of lawyers use some AI tool in daily work — mainly for case/literature research, document analysis, contract drafting, and routine automation. 62% reportedly noted weekly time savings of 6–20%.
• Volume: legal AI Harvey and Thomson Reuters' CoCounsel reported processing over 10 million legal documents (contracts, court filings, regulatory submissions, discovery materials, patents) in Q1 2026 alone.
• Accounting/tax adoption: generative-AI use at tax, accounting, and audit firms jumped from 8% in 2024 to 21% in 2025 (the biggest increase among surveyed industries).
• Employment shift: a Stanford study found that in AI-exposed fields like accounting, early-career jobs fell 13% versus 2022, while experienced professionals stayed stable to growing. Accountants showed +5% growth; bookkeepers, -5%.

In short, the "research, draft, and record" work is being replaced by AI, and professional value is moving to judgment and responsibility. The question has shifted from "use it or not" to "how far do you delegate, and what do humans keep their hands on?" From here, we look at the breakdown.

2. The work AI changes (by profession)

"The professions" cover a lot, and the key points differ by role. Let's look at three representative ones. What they share is the structure "AI does the groundwork (research, drafting, recording); a human makes the final judgment."

What's notable is that the efficiency creates "new room." General-purpose AI like data analysis, minutes, and turning documents into text (OCR) also directly help with professional groundwork. Redirecting the time freed from routine work into strategic proposals for clients and intricate judgments — that's the common trait of professionals who are growing now.

3. The biggest pitfall: fabricated case citations

Behind the efficiency, AI use in the professions carries a fatal risk. As the opening example shows, AI plausibly "invents" non-existent cases, statutes, and sources (hallucination). In law and accounting, that connects directly to instant loss of trust and disciplinary action.

The industry's answer is clear: "don't use unverified output." Specialized tools focus on exactly this. Harvey, for instance, touts 99.7% verified-citation accuracy and a policy of not including citations it can't verify with confidence — it flags them instead. CoCounsel returns citations grounded in a case database (Westlaw), suppressing the invention problem by "only ever citing cases that actually exist." Using plain ChatGPT as-is versus using these verification-backed specialist tools is a different order of risk. Here, the line between what AI can and can't do can literally affect your professional life.

4. What does NOT change — the essential value

We've seen "what changes." But what truly matters is "what doesn't change." There are areas where delegating to AI causes accidents, or that can't be fully delegated at all.

Here's the crux. AI makes "plausible answers" fast and in bulk. But AI has no professional skepticism and no ethical judgment, and it answers confidently even when wrong. Gray judgments decided by intent and context — like "is this hobby income or business income?" — are exactly where AI is weakest. And the decisive one is "responsibility": an audit opinion and legal advice are, in the end, signed by someone who takes on legal liability. That responsibility can't be handed off to AI. Which is why, in the AI era, a professional's value converges not on "speed of work" but on "being able to take on judgment and responsibility." On the broader shift, see also jobs that survive the AI era.

5. The junior crisis and new roles

The shadow falls heaviest on juniors. As in the Stanford study above, automating the "apprenticeship routine" of research and bookkeeping narrows the entry point for the inexperienced. The figures of fewer bookkeepers and more accountants are a microcosm. And a serious question arises — if there are fewer chances to build skills through grunt work, how will the next experts be trained?

At the same time, AI is creating new roles. "AI compliance officers" who ensure a firm's AI use is ethical, transparent, and audit-ready; "tax prompt engineers" who extract precise legal and tax insights from AI — the expertise of wielding AI correctly is itself becoming a job. What newcomers are asked for is not "speed of work" but the ability to use AI as a tool from day one, verify its output, and turn it into judgment. This overlaps with the debate over whether veterans or juniors are more at risk — the AI era's homework for every industry is right here.

6. Advice by role

So how should you act, by role? Let's translate the survey into practice.

Especially important is handling confidentiality. Entering case records, financial information, or personal data into a cloud AI means sending it externally. For professionals bound by confidentiality, checking the data handling policy and choosing a design/plan that doesn't use your input for training is essential. The instinct for what not to enter into AI is demanded even more strictly in the professions.

Summary

Here's AI's impact on the legal, accounting, and tax professions, sorted out.

• The work transforms: AI speeds up research, contract review, bookkeeping, and draft returns. Over 90% of lawyers use it; accounting-firm adoption rose 8% → 21%.
• The shared structure: AI does the groundwork; humans make the final judgment, sign, and bear legal liability. Common to all three professions.
• The biggest risk: hallucination, like fabricated case citations. Verification-backed specialist tools plus checking primary sources are essential.
• The unchanging core: professional skepticism, ethics, gray judgments, and responsibility stay with humans. AI can't sign.
• Watch out for juniors: apprenticeship routine work is automating. But AI-use roles are also being born.
• Clients get smarter too: pay for "judgment and responsibility," not the work. Confirm how confidentiality is handled.

In the end, what AI poses to the professions is the question: "is what you sell the work, or the judgment and responsibility?" In an era when AI drafts research, contract reviews, and tax returns in seconds, what a client entrusts to a professional is the resolve to judge correctly in intricate situations, uphold ethics, and write their name to take on the responsibility. That essence won't waver for a while, no matter how smart AI gets.

FAQ