I interviewed six attorneys whose work covers litigation, patent law, construction disputes, commercial law, government investigations and business transactions. Each one answered the same ten questions about how AI is changing legal work today and how it may change the profession over the next ten years.
All six already use AI in their practice. Their answers fall into four parts:
Where the attorneys agree
On three points, every attorney gave the same answer: AI is already part of legal work, it is best at high-volume document tasks, and its work must always be checked.
AI is already a standard legal tool
How often legal professionals use AI (2026)
Percentage of legal professionals at each level of use
Source: Thomson Reuters Institute, 2026 AI in Professional Services Report (February 2026). In total, 81% use AI at least weekly. A separate 2026 survey by SurePoint found that 63% of mid-sized U.S. law firms have formally adopted generative AI.
AI allows us to be rounding second base, not standing in the batter's box waiting for a pitch.
In other words, AI handles the slow early stages of a case, such as research that leads nowhere, so lawyers can move directly to the strongest cases and arguments. His firm now trains every lawyer, paralegal and legal assistant to use it.
AI is best at high-volume document work
Every attorney named the same strength: AI can read, sort and summarize large amounts of material very quickly. That work used to belong to junior lawyers and paralegals.
Share of AI-using legal professionals who use it for each task. Source: Thomson Reuters Institute, February 2026.
It's a digital first-year associate. It gets you 80 percent of the way. It's not perfect.
A lawyer must always check AI's work
Every attorney raised this warning without being asked. AI tools sometimes "hallucinate," meaning they state false information, including court cases that do not exist, with complete confidence. Research confirms that this happens often.
How often AI tools make up false legal information
Hallucination rate on legal research questions
Source: Stanford RegLab / HAI, study published 2024; peer-reviewed version 2025. Even tools built specifically for lawyers gave false answers at least one time in six.
Legal professionals' concerns about AI are growing
Percentage of legal professionals who share each concern
Source: Thomson Reuters Institute, 2025 and 2026 reports. Reliability remains the top worry: in a survey of law firm leaders released in March 2026, 81% reported concern about AI's reliability and risk (SurePoint).
Where their predictions split
When the questions turned to the next ten years, the attorneys' answers differed on four issues: how many lawyers will be needed, which areas of law will change most, what AI means for people without lawyers, and whether the legal system will be better off.
Will there be fewer lawyers?
Yes, significantly fewer
If AI does the work of junior lawyers, firms will hire fewer of them. One litigator predicted that a legal team of five to seven lawyers on a major case could shrink to three or four.
"We will not need as many lawyers coming out of law school."
Partner, trial & appellate group, Jackson Walker20 years; government investigations
No, demand is growing
Others argue that AI is creating new legal work. Clients need help understanding the laws and regulations around AI itself, and retiring lawyers must still be replaced.
"In 2026, AI has only increased the demand for legal services."
Commercial attorney, Duane MorrisBusiness-to-business disputes in the energy and health care industries
Which areas of law will change the most?
The attorneys named different areas, but a pattern appeared: the more routine and lower-stakes the work, the more exposed it is. Several litigators, however, argued that both contract work and litigation require too much judgment for any whole field to be replaced.
Most affected
- Routine, low-stakes matters such as traffic tickets and minor misdemeanors
- Real estate documents, wills and trusts
- Standard contracts, employment agreements and waivers
- First-pass document review
- Much of family law
Least affected
- Arguing a client's case in court
- High-stakes matters that require decades of experience
- Criminal law, especially prosecution
- Strategy, negotiation and complex decisions
- Judging, especially appeals
Replace judges in the next ten years? No way. Maybe in 500.
You will always need a human prosecutor.
Is AI helping people who represent themselves in court?
Three attorneys raised an effect I did not expect. People who cannot afford a lawyer, called pro se litigants, now use chatbots to write their own court filings. The attorneys disagreed on whether this is progress or a problem.
Court records show how quickly this has grown. A public database maintained by researcher Damien Charlotin at HEC Paris tracks every court decision it can find in which a filing contained false, AI-generated material, such as invented cases or fake quotations. The first well-known case was in New York in June 2023. By September 2026, the database listed 2,046 cases. About two-thirds of them (1,397) were in U.S. courts, and the rest came from more than 30 other countries, led by Canada, Australia and the United Kingdom. Most were filed by people without lawyers.
Court cases caught with fake, AI-invented citations
Running total of court cases where a filing cited made-up cases or quotes created by AI, May 2025 to Sept. 2026
Sources: Charlotin database counts as reported by Reason/Volokh Conspiracy (May 2025), PlatinumIDS (April 2026) and the HAQQ sanctions tracker (May to Sept. 2026). In U.S. courts alone, 487 such errors were documented in 2025, more than ten times the 2024 total (SurePoint, March 2026).
Who submitted the fake, AI-invented citations?
All 2,046 court cases where a filing contained made-up legal sources, June 2023 to Sept. 21, 2026, by who filed it
Source: Damien Charlotin, AI Hallucination Cases Database, accessed September 22, 2026. Some cases involve more than one party.
Five U.S. cases where lawyers were caught filing fake, AI-invented citations
In each case, a lawyer filed court papers that relied on AI and included legal sources that did not exist. The judge discovered the errors and punished the lawyers responsible.
-
June 2023Mata v. AviancaFederal court, New York City
- The lawsuit
- A passenger sued the airline Avianca, saying a metal serving cart injured his knee during a flight.
- What went wrong
- His lawyers used ChatGPT for research. Their brief cited six court cases that ChatGPT had invented, complete with fake quotations.
Penalty$5,000 fine for the two lawyers and their firm. This was the first widely reported case of its kind. -
February 2025Wadsworth v. WalmartFederal court, Wyoming
- The lawsuit
- A family sued Walmart, saying a hoverboard bought there caught fire and burned their home.
- What went wrong
- A lawyer used his law firm's own AI tool to draft a motion. Eight of the nine cases it cited did not exist.
Penalty$5,000 in combined fines for three lawyers. The lawyer who drafted it was removed from the case. -
July 2025Coomer v. LindellFederal court, Colorado
- The lawsuit
- A former voting-machine company employee sued MyPillow founder Mike Lindell for defamation.
- What went wrong
- Lindell's lawyers filed a brief with nearly 30 defective citations, including cases that did not exist. One lawyer admitted using AI.
Penalty$3,000 fine for each of the two lawyers. -
December 2025Couvrette v. WisnovskyFederal court, Oregon
- The lawsuit
- A family dispute over the ownership of a winery, which also included claims of elder abuse.
- What went wrong
- The plaintiffs' lawyers filed three briefs containing 15 cases that did not exist and 8 made-up quotations.
PenaltyAbout $110,000 in fines and legal fees, the largest AI penalty reported so far. The judge also threw out the plaintiffs' entire case. -
2026Withers v. City of AberdeenFederal court, Northern Mississippi
- The lawsuit
- An attorney sued the city of Aberdeen, Mississippi, over unpaid bills for his legal work.
- What went wrong
- Lawyers on both sides used AI to draft filings and did not check the cases it cited, several of which were fake.
PenaltyAll four lawyers were removed from the case and fined. Two were banned from that federal court for two years, and the trial was canceled.
Sources: court decisions as reported by LawSites, Mondaq, FindLaw, Bloomberg Law and the HAQQ and GC AI sanctions trackers (2025–2026).
AI has made the courthouse far more accessible.
Pro se filings have flooded the courts.
A third attorney noted that responding to these filings adds work for lawyers on the other side. Still, the partner concerned about the courts believes AI will improve the system overall, because it can make legal help affordable for people who could never pay for a lawyer.
Will the legal system be better, worse, or simply different?
I ended each interview with this question. All six attorneys answered it, and none of them said "worse."
It's like a gun. It can be used for good and bad.
What this means for future lawyers
The work AI does best is the work that new lawyers have always started with. That affects how many graduates firms hire, how new lawyers are trained, and how students should prepare.
Law firms are hiring fewer new graduates
Hiring data shows a gradual decline. Firms are offering fewer summer positions, the largest firms have not grown their first-year classes since 2023, and fewer graduates found law firm jobs in 2025 than in 2024.
AI is one likely cause among several. Clients have pushed back on legal costs, firms increasingly hire experienced lawyers from other firms, and companies are keeping more legal work in-house. However, industry analysts note that firms were using AI widely by 2025, so it is reasonable to conclude that AI now plays a role.
Law firms hired fewer new graduates in 2025
New law graduates hired by law firms, Classes of 2022 to 2025
Source: NALP employment reports for the Classes of 2022 through 2025 (Class of 2025 data released September 2026, the most recent available). The 2022 figure is estimated from NALP's reported change for 2023. Law firm hiring peaked with the Class of 2024, which was unusually large; the Class of 2025 was about 7% smaller, which accounts for part of the decline.
First-year associates hired by the 200 largest U.S. firms
Hiring peaked in 2023 and has remained below that level.
Source: SurePoint Law School Hiring Report, published July 2026 (2025 is the latest full year). The vertical axis starts at 7,300 so the change is visible. Summer programs are also shrinking: the average summer class fell from 14 to 12, and the median office made only 6 offers for 2025, the lowest NALP has recorded.
How will new lawyers learn?
When asked what concerned them most, the attorneys mentioned false information and confidentiality. The concern they returned to most often, however, was that new lawyers might stop thinking for themselves.
One partner has seen associates submit AI-generated work without understanding it. Another warned that young lawyers face a strong temptation to accept AI's answers without question. A third explained that routine tasks are how lawyers build and maintain their skills. If AI performs those tasks, firms will need a new way to train beginners.
There is value in inefficiency.
People are not learning to think.
The attorneys' advice for students considering law school
None of the attorneys advised against going to law school. All of them, however, said the path into the profession is changing.
Aim for a top-20 law school
With firms hiring fewer new lawyers, she advised attending one of the 20 highest-ranked law schools in the country, often called a "T20" school. Graduates of lower-ranked schools may struggle to find jobs. If law is truly your passion, pursue it, but build a strong résumé and be prepared to compete.
Learn to use AI now
The lawyers who succeed will be those who use AI to become more valuable. Writing and advocacy remain essential, but AI skills are now equally important.
Do your own thinking
Never submit work you do not understand. Know what a document says and what it means. Law school is valuable because it teaches you how to think.
Build skills without shortcuts
New lawyers should not rely on AI at first. Working through problems on your own is how you develop judgment, and routine work keeps your skills sharp.
Prepare for competition
With fewer lawyers needed on each case, entry-level positions will be harder to get, and graduates may compete with experienced lawyers for the same jobs.
The profession will still need you
Experienced lawyers retire and must be replaced, so firms will continue to develop new talent. He would not hesitate to enter the profession today, and he recommends thinking about AI early.
What it all adds up to
Taken together, the interviews and the data point to four conclusions.
The change is real, but it is gradual
By 2026, 81% of legal professionals used AI at least weekly, and known court cases involving AI errors grew from 87 in May 2025 to more than 2,000 in September 2026. Hiring, however, is declining slowly rather than collapsing. The attorneys agree on what AI can do today. They disagree about how far and how fast it will go.
Entry-level work is changing first
Document review, first drafts and routine matters are where AI performs best, and those tasks have traditionally belonged to new lawyers. Both the interviews and the hiring numbers show the effects starting at the bottom of the profession.
The profession faces a training problem
The skills the attorneys believe AI cannot replace, such as judgment, strategy and advocacy, were developed through years of the same routine work that AI now performs. Law firms and law schools will need new ways to train the next generation.
The largest effect may be outside law firms
People without lawyers are already using AI to go to court on their own. This could make the legal system more accessible, but it also places new strain on courts. How courts respond may matter as much as how law firms adapt.
Not according to any of the attorneys I interviewed. But it will change what lawyers do.
AI is taking over much of the work that once filled a lawyer's day, and the profession will likely need fewer people, especially at the entry level. The lawyers who remain will be valued for what AI cannot provide: sound judgment, persuasive advocacy, empathy for clients and the ability to recognize when the AI is wrong.
For students considering law school, the message is not to stay away. It is to prepare differently: aim for a top-ranked school, learn to use AI well, and never let it replace your own thinking.
The attorneys behind the answers
Thank you to the six attorneys who shared their time and insight.
- Practice
- Transactional law
- Expertise
- Private offerings, contract drafting and review, AI law
- Practice
- Litigation and defense
- Expertise
- Personal injury, commercial construction defects, wills and powers of attorney
- Practice
- Commercial law, business-to-business disputes
- Expertise
- Energy and health care industries
- Practice
- Trial and appellate, 20 years
- Expertise
- Government investigations and responding to complaints for companies and individuals
- Practice
- Commercial and construction litigation, 24 years
- Expertise
- Represents property owners in large construction disputes, working with experts to identify defects and directing case strategy and management through litigation
- Practice
- Patent litigation and licensing, more than 25 years
- Expertise
- Patent enforcement in federal courts, the U.S. International Trade Commission and the Federal Circuit; patent licensing negotiations
- Thomson Reuters Institute, 2026 AI in Professional Services Report (legal highlights, via NY Daily Record)
- Thomson Reuters Institute, 2026 AI in Professional Services Report
- SurePoint, 2025 State of the Legal Industry Report, March 2026 (via LawSites)
- Stanford HAI, "AI on Trial: Legal Models Hallucinate in 1 out of 6 (or More) Benchmarking Queries"
- Damien Charlotin, AI Hallucination Cases Database (accessed Sept. 22, 2026)
- Eugene Volokh, "AI Hallucination Cases," Reason, May 18, 2025
- PlatinumIDS, "1,227 Fabricated Citations and Counting," April 2, 2026
- HAQQ, AI Hallucination Sanctions Tracker (updated Sept. 21, 2026)
- GC AI, AI Hallucination Legal Cases: A Sanctions Tracker (2026)
- NALP Class of 2025 employment data (via National Jurist, Sept. 2026)
- SurePoint Law School Hiring Report 2026 (via Artificial Lawyer, July 2026)
- NALP summer associate recruiting data (via JD Journal, March 2026)