Lawyer / Attorney
Lawyers face heavy AI exposure in the document-and-research-heavy parts of the job — a structural exposure score of just 14 out of 100, among the lowest 'safe' readings in this dataset, reflecting how much of legal research, contract review, and first-draft writing overlaps with what large language models already do well.
We have not ingested real-world usage data for this occupation yet. We show a band only where genuine data exists, rather than estimate one.
What a lawyer / attorney actually does
Lawyers advise clients on their legal rights and obligations, draft and negotiate contracts and other legal documents, and represent clients in court, arbitration, or before regulatory bodies. Day-to-day work spans a wide range depending on practice area — a litigator spends much of the week on legal research, drafting motions, and preparing witnesses for trial; a transactional attorney negotiates deal terms and drafts contracts; an in-house counsel advises a company on regulatory compliance and risk.
What unites all of it is that a lawyer's work product carries personal legal weight: when an attorney signs a court filing, it's a certification under rules like Federal Rule of Civil Procedure 11 that they've personally reviewed it and believe it's warranted by the facts and law, and when they give legal advice, the client is relying on the attorney's professional judgment and license, not just the accuracy of the underlying research.
A large share of the underlying tasks — legal research, contract review, document summarization, first-draft memo writing — is squarely within reach of large language models, and legal-tech funding has surged accordingly as firms adopt AI research and drafting assistants. But the license to practice law, the duty a lawyer owes a client, and the legal accountability for what gets filed or advised remain attached to a specific, bar-admitted human, not to a tool.
Why it reads this way
Lawyers face heavy AI exposure in the document-and-research-heavy parts of the job — a structural exposure score of just 14 out of 100, among the lowest 'safe' readings in this dataset, reflecting how much of legal research, contract review, and first-draft writing overlaps with what large language models already do well. What keeps the role AI-resistant isn't the difficulty of the underlying tasks, it's the legal wall around who's allowed to do them: unauthorized-practice-of-law statutes in every U. S.
state make it illegal for a non-lawyer (and, by clear extension, an AI system acting autonomously) to represent a client in court, give legal advice, or sign a court filing. Only a bar-admitted attorney can do those things, and only that attorney is personally liable for malpractice and subject to professional discipline if the work is wrong. Mata v.
Avianca (2023) — where lawyers were sanctioned for submitting a ChatGPT-hallucinated brief citing cases that didn't exist — became the case that crystallized this for the profession, and it directly prompted the American Bar Association's first formal ethics guidance (July 2024) stating attorneys cannot rely on AI-generated legal work without independently verifying it. Legal-tech investment has surged (roughly $3.
56 billion in the first half of 2025 alone), but industry coverage of that spending consistently frames it as building tools that produce 'verifiable, trustworthy outputs legal teams can rely on' — an assistance model, not an autonomy model. That combination of heavy task-level AI use alongside a hard, licensed line around who can actually practice law is why we rate this occupation 'AI-resistant despite the paperwork' rather than either a clean 'safe' or an 'at risk' verdict.
Skills this role draws on
We don't yet have task-by-task time-share data for this occupation, so we can't show which specific tasks carry the exposure score above. This is the real skill set instead.
Typical salary range (USD)
$78,360–$351,600
Average: $159,670/yr
Range source: BLS OEWS wage data via O*NET OnLine (2025 release), SOC 23-1011.00 'Lawyers'
Average source: BLS OEWS wage data for Lawyers (23-1011.00) via O*NET OnLine, 2025 wage data (Annual Median Wage)
Related careers
Head-to-head comparisons
Lawyers score a structuralScore of just 14/100 — among the lowest 'safe' readings in this dataset — reflecting how much legal research, contract review, and first-draft writing overlaps with current LLM capability. What protects the role is unauthorized-practice-of-law statutes that make representing a client, giving legal advice, or signing a court filing a licensed, personally-liable act a non-lawyer (or an autonomous AI system) cannot legally perform. Mata v. Avianca (2023) — sanctions for a ChatGPT-hallucinated brief — directly prompted the ABA's July 2024 ethics guidance that lawyers cannot rely on AI output without independent verification. As of 2026-08-11, legal-tech investment (~$3.56B in H1 2025) is real but consistently framed around producing verifiable outputs for licensed lawyers, not replacing the license itself.
This is a researched judgment (checklist v1.2), not a statistical measurement — it sits alongside, not instead of, the structural exposure signal above. Next scheduled review: 2027-02-11.
Data sources & methodology
Salary data: BLS OEWS wage data via O*NET OnLine (2025 release), SOC 23-1011.00 'Lawyers'. Min/max figures represent the 10th–90th percentile annual wage range across the United States. Average figure sourced separately: BLS OEWS wage data for Lawyers (23-1011.00) via O*NET OnLine, 2025 wage data (Annual Median Wage).
Automation Risk Score: Based on O*NET occupational analysis (23-1011.00) evaluating task complexity, physical requirements, social intelligence, and environmental variability. Methodology based on research from Frey & Osborne (Oxford, 2017).
Growth projections: 3-4% (2024-2034), about as fast as the all-occupation average (O*NET OnLine / BLS OEWS projections), based on BLS Occupational Outlook Handbook.