Generative AI reached law offices faster than the rules governing it, and courts have spent the last few years catching up. Judges are no longer treating AI use by attorneys as a hypothetical. They’re setting expectations around disclosure and making clear that lawyers remain responsible for the work they submit, including the accuracy of their citations.
For an employment litigator such as Emily Armstrong Hinsdale, who divides her work between an Arizona firm and clients reached from her base in DuPage County, Illinois, those questions are part of everyday legal practice. Her own account of building a practice around workplace disputes describes work centered on research, drafting, and documentation — areas where generative AI can easily enter the process.
Disclosure may be the newer question, but the underlying obligations aren’t. Candor and professional responsibility applied to legal work long before generative AI became part of it.
Why Courts Started Asking the Question
A major turning point came in 2023, when attorneys in a New York federal case submitted a filing containing case citations that didn’t exist. The authorities appeared plausible, complete with reporter information, but fell apart when opposing counsel tried to locate them. The sanctions that followed drew widespread attention, and judges began responding with their own requirements.
Federal judges started issuing standing orders addressing generative AI. In the Northern District of Illinois, for example, a magistrate judge required parties using generative AI to prepare or draft filings to disclose that use and identify the tool involved. The order also emphasized that Rule 11 of the Federal Rules of Civil Procedure hadn’t changed. An attorney’s signature still certifies that a reasonable inquiry has been made into the law and facts presented to the court.
Existing professional obligations already give courts a framework for addressing AI-generated errors. A lawyer signing a brief is responsible for making sure cited authorities exist and support the propositions attributed to them. Generative AI doesn’t change that responsibility. It adds another place in the drafting process where something can go wrong.
What Disclosure Actually Requires
Disclosure requirements aren’t uniform. Some judges require a statement when generative AI has been used in preparing a filing. Others may require broader certifications or disclosures. Certain courts have also imposed restrictions on how generative AI can be used, sometimes distinguishing those tools from AI-assisted features already built into established legal research platforms.
Drawing a clean line around “AI” isn’t always easy. Ranking, recommendation, and natural-language search systems have existed within commercial research databases for years. Rules focused specifically on generative drafting or research can therefore be more precise than requirements that attempt to cover every technology incorporating some form of artificial intelligence.
Professional responsibility guidance has developed alongside the court rules. The American Bar Association’s Formal Opinion 512 addresses generative AI through familiar ethical duties, including competence, confidentiality, communication, and reasonable fees.
The technology may be new, but much of what lawyers are being asked to do isn’t: understand the tools they’re using, protect client information, and independently verify their work.
Why Emily Armstrong Hinsdale Sees Verification as the Real Obligation
Disclosure tends to get more attention, but it doesn’t replace verification. Telling a court that AI helped prepare a filing doesn’t excuse inaccurate or invented authority. Even when a particular court doesn’t require disclosure, attorneys are still responsible for what they submit.
Lawyers who practice across jurisdictions face an additional complication because the requirements can change from one courtroom to the next. Practitioners like Emily Armstrong Hinsdale, whose labor and employment practice spans an Arizona firm and matters handled from Illinois, may encounter different expectations depending on the court and assigned judge.
Checking local rules and standing orders has therefore become another part of preparing a filing. AI requirements can be handled much like page limits, formatting rules, and other judge-specific procedures: check them before the document reaches the filing stage.
Armstrong’s background in industrial engineering also shapes how she approaches process. In legal drafting, a reliable verification system shouldn’t depend on someone remembering at the last minute that citations still need to be checked. Building that step into the workflow makes it part of preparing the document in the first place.
The Confidentiality Problem Nobody Puts in the Certification
Accuracy isn’t the only concern. Generative AI can also create questions about client confidentiality.
Employment matters often involve sensitive information, including personnel records, medical documentation related to leave requests, internal investigation materials, settlement terms, and the identities of employees who report misconduct. Entering that information into an external AI system without knowing how the data is handled can create problems that have nothing to do with the accuracy of the eventual filing.
Attorneys with experience handling trade secret and non-competition claims, work that was part of Armstrong’s early career at a global firm’s Boston office, already know how important it is to control where sensitive information goes. Once confidential material leaves a controlled environment, it may be difficult to undo that disclosure.
Before using a tool with client information, attorneys need to know whether it retains prompts or other inputs, what contractual protections apply, and whether that information may be used for training or other purposes. Sensitive identifiers can be removed when appropriate, and written firm policies can give associates and staff clear guidance instead of leaving those decisions to each individual user.
Building a Verification Habit That Holds Up
Verification doesn’t need to become a separate, complicated process. It works better when it’s part of the normal one.
Before drafting begins, attorneys can check the assigned judge’s standing orders and applicable local rules. Any citation generated or suggested by an AI tool should remain unverified until someone confirms it through an authoritative legal source, checks that the case remains good law, and makes sure it supports the proposition in the draft.
Firms can also keep internal records of which tools were used in preparing filings. If a court later asks for disclosure, the answer can come from documentation instead of memory.
Responsibility should be clear as well. Assigning a particular attorney to verify citations and compliance on each matter reduces the chance that everyone assumes someone else completed the review. Associates, staff, and contract attorneys should be working from the same internal policies.
For employment lawyers, there’s something familiar about this kind of recordkeeping. It’s similar to the documentation practices they routinely recommend to clients, applied inside the firm. If a process is later questioned, clear records make it much easier to establish what happened and how the work was handled.
Where This Is Heading
Courts and professional organizations continue to develop guidance around generative AI. Some jurisdictions have moved beyond individual standing orders toward broader policies, while others still leave requirements to individual judges.
For attorneys practicing in multiple courts, the immediate reality is that the rules still need to be checked matter by matter. They may become more consistent over time, but lawyers can’t assume that has happened yet.
Whatever the rules of a particular court, the attorney’s responsibility for the filing remains. Generative AI may change how some of the research and drafting gets done, but it doesn’t remove the need to verify the work, protect confidential information, or know the requirements of the court where that work will be submitted.
About Emily Armstrong
Armstrong founded the Law Offices of Emily Armstrong, LLC in November 2015, a Phoenix practice devoted to labor and employment litigation that she now directs remotely from Hinsdale in DuPage County, Illinois. She also serves as of counsel to Matheson & Matheson, PLC in Scottsdale and is a member of the Arizona Employment Lawyers Association, the state affiliate of the National Employment Lawyers Association.
Her earlier positions included associate roles at Milligan Lawless, P.C. in Phoenix and Nixon Peabody LLP in Boston, and she writes occasionally about changes reshaping labor and employment practice.
Emily D. Armstrong is an employment lawyer based in DuPage County, Illinois, and a member of the Arizona Employment Lawyers Association. She owns a law firm headquartered in Arizona, which she runs remotely from her home in Illinois, and she earned her Juris Doctor from the DePaul University College of Law after completing a bachelor’s degree in industrial engineering at the University of Missouri-Columbia. A supporter of numerous nonprofit organizations and a proud mother of two, she makes time for tennis and travel when her schedule allows.
Where the legal industry reads first.
Enjoyed this article? Get the biggest legal industry updates, deals, appointments, insights and expert interviews in your inbox, free.
No spam. Unsubscribe anytime.From Legal Desire
Is your firm the one being cited, or the one being skipped?
We ran the test. On several everyday legal questions, software companies are answering and law firms are not. We help firms fix that, and we build the technology behind the practice.
