AI in Legal Services: Where Law Firms Are Drawing the Line

AI in legal services refers to the application of artificial intelligence tools across the practice of law — including contract analysis, legal research, document drafting, due diligence, compliance monitoring, and litigation support. Law firms and in-house legal teams are adopting AI at a pace that has accelerated significantly since 2024, but the legal domain presents a governance challenge that few other industries match: the consequences of AI errors can be immediate, costly, and in some cases, career-ending. Human Agency works with legal teams and regulated-industry organizations to build AI programs where the efficiency gains are real and the governance is built in from the start.

Why legal AI has reached a turning point

Legal work has always been information-intensive, time-consuming, and expensive to scale. A senior associate reviewing a 300-page contract for non-standard clauses, a paralegal researching precedent across thousands of cases, a compliance team monitoring regulatory updates across dozens of jurisdictions — these are the tasks that define how legal departments spend their time and how law firms bill their hours.

The scale of AI adoption in the legal sector is no longer speculative. According to the Thomson Reuters 2025 Future of Professionals Report — which surveyed 2,275 professionals across more than 50 countries — 80% of law firm professionals expect AI to fundamentally transform how they conduct business. Usage is already following: the Thomson Reuters Institute’s 2025 Generative AI in Professional Services Report found that AI adoption in legal organizations nearly doubled year-over-year, from 14% in 2024 to 26% in 2025. The American Bar Association’s 2024 Legal Technology Survey found that 30% of lawyers were already using AI tools, nearly triple the 11% who did so in 2023.

What changed is not that the tools became available — early legal AI products have existed for years. What changed is that the quality of outputs crossed a threshold where the time saved is real, the error rate is manageable, and the competitive pressure to adopt is visible. A firm reviewing contracts in four hours while a peer firm does the same work in forty-five minutes is not just faster — it is cheaper for the client, creating commercial pressure that every firm in the market now feels.

Where AI creates genuine value in legal practice

The legal tasks where AI produces the most reliable value share a common structure: they are high-volume, pattern-dependent, and require scanning large amounts of information to surface what is relevant. These are the tasks that consume associate and paralegal hours at scale.

Contract review and analysis

AI contract review tools scan agreements for non-standard clauses, flag deviations from approved templates, identify missing provisions, and surface risk factors — at a fraction of the time a human reviewer would need. In M&A due diligence, where thousands of contracts may need review in a compressed timeline, the efficiency case is clear and well-documented. AI contract review platforms such as Luminance, Ironclad, and tools integrated into established legal technology vendors are among the options firms and legal operations teams are evaluating.

The governance line is important: AI surfaces issues for human review; it does not make the legal judgment about whether a clause is acceptable. The attorney still owns that decision. AI that flags a non-standard indemnification clause is a drafting assistant. AI that advises whether to accept the clause based on the client’s risk profile is providing legal analysis — and that requires a licensed attorney, not just a model.

Legal research

AI-assisted legal research tools can scan case law, statutes, regulations, and secondary sources far faster than any human, surface relevant precedent across jurisdictions, and synthesize patterns in how courts have ruled on specific issues. Westlaw’s AI features and LexisNexis Lexis+ AI have become common in larger practices.

The critical risk in AI legal research is hallucination. Language models that generate citations that do not exist — a problem that has already resulted in court sanctions for attorneys who submitted AI-generated briefs without verification — remain a real concern. No AI-generated legal citation should ever be filed or relied upon without human verification against the actual source. The Mata v. Avianca case (SDNY, 2023) is the landmark public example: attorneys were sanctioned after submitting fabricated case citations generated by ChatGPT. This is not a theoretical caution.

Document drafting and templates

AI tools that draft routine legal documents — NDAs, service agreements, employment contracts, standard corporate filings — from approved templates and structured inputs save significant time on work that does not require creative legal judgment. The value is highest for high-volume, standardized documents where the goal is speed and consistency rather than bespoke legal analysis.

Compliance monitoring

In-house legal and compliance teams monitoring regulatory changes across multiple jurisdictions have adopted AI monitoring tools that track regulatory publications, flag relevant changes, and summarize implications — reducing the manual scanning that previously consumed significant compliance analyst time.

Where legal AI governance is non-negotiable

The legal domain has governance requirements that do not apply in the same way to most enterprise AI contexts. Several are worth naming explicitly before any AI deployment in a legal function.

• Attorney-client privilege. AI tools that process privileged communications must be evaluated for whether their use constitutes waiver, and the data handling requirements of any AI vendor must be assessed against privilege obligations.
• Confidentiality obligations. Rule 1.6 of the Model Rules of Professional Conduct requires attorneys to protect client information. Using AI tools that send client data to third-party servers without appropriate data processing agreements creates confidentiality exposure.
• Competence obligations. Rule 1.1 requires attorneys to maintain competence, which bar associations are increasingly interpreting to include understanding the AI tools used in practice. California, Florida, and New York have all issued formal guidance on AI use in legal practice.
• Supervision obligations. Attorneys remain responsible for work product regardless of how it was produced. AI-generated content that is submitted without attorney review is still the attorney’s work product for purposes of professional responsibility.

These obligations are not reasons to avoid AI in legal contexts. They are governance design requirements that must be satisfied before AI is deployed — not discovered afterward.

What separates the law firms getting this right

The firms and legal departments seeing the best results from AI in 2026 share a consistent pattern: they defined clear boundaries between AI-assisted tasks and attorney judgment tasks before deployment, built governance into the workflow rather than adding it afterward, and invested in enabling their attorneys and staff to use AI confidently — not just gave them access.

The firms struggling are the ones that deployed tools without the governance conversation, or that deployed in ways that eroded attorney trust by producing outputs that required more cleanup than the tool saved. The technology is the easier problem. The workflow design and governance are where most legal AI programs succeed or fail. This mirrors what Human Agency observes across every regulated industry: the organizations that treat AI adoption as a people-and-process change, not just a technology deployment, see adoption that sticks.

Frequently Asked Questions

How are law firms using AI in 2025 and 2026?

Law firms are using AI primarily in four areas: contract review and analysis, legal research and precedent identification, document drafting from approved templates, and compliance monitoring across regulatory sources. The ABA’s 2024 Legal Technology Survey found that 30% of lawyers were already using AI tools — nearly triple the 11% who did so in 2023 — with adoption highest in firms of 100 or more attorneys, where nearly half reported AI use. The pattern across firms doing this well is consistent: AI handles the high-volume, pattern-dependent tasks, and attorneys own the legal judgment calls.

What are the biggest AI risks for legal teams?

The three most significant risks are hallucination in legal research — AI-generated citations that do not exist, which have already resulted in documented court sanctions — confidentiality exposure from client data processed by AI tools without appropriate security controls and data processing agreements, and competence risk from attorneys submitting AI-generated work product without adequate review. Human Agency designs AI governance frameworks for legal and regulated-industry organizations that address all three before tools go into use, not after something goes wrong.

Do professional responsibility rules restrict how attorneys can use AI?

Yes, in ways that are still evolving. The duty of competence under Rule 1.1 is increasingly interpreted by state bars to require understanding the AI tools used in practice. The duty of confidentiality under Rule 1.6 requires that attorney-client information be protected regardless of what tools are used to process it. The duty of supervision means attorneys remain responsible for AI-generated work product. California, Florida, and New York have all issued formal guidance on AI in legal practice, and the ABA has published formal opinions addressing AI use. Any legal team deploying AI should review the guidance issued by the relevant state bars before deployment.

How do legal teams get started with AI without creating compliance risk?

Start with the tasks that are genuinely lower-risk: internal research, first-draft contract templates for standard agreements, compliance monitoring summaries for internal use. Build the governance infrastructure — data processing agreements with vendors, confidentiality protocols, attorney review requirements, verification procedures for any AI-generated research — before expanding to client-facing or filing-critical work. Human Agency works with legal teams and regulated-industry organizations on AI readiness assessments and governance frameworks that satisfy professional responsibility requirements while enabling the efficiency gains that make AI worth deploying.