In April 2026, the Ohio Board of Professional Conduct published its Ethics Guide, Artificial Intelligence for Lawyers and Judicial Officers. It is nonbinding staff guidance, not a formal opinion or a rule amendment, but every Ohio practitioner should treat it as the clearest statement yet of how the existing Rules of Professional Conduct apply to the tools many of us already use every day.
The headline is reassuring. The Ohio Rules of Professional Conduct do not prohibit lawyers from using AI to assist in providing legal services. The catch, and there is always a catch, is that the same duties that have always governed your practice now extend squarely to these tools. Nothing about the technology lowers the bar. If anything, it raises the stakes for the lawyer who does not understand what is happening under the hood.
Here is what the Guide actually requires, and what it means for how you run your practice.
Competence now includes the machine
The Guide places AI in the same category as email, internet research, cloud storage, and the smartphone. Each became, over time, a relevant technology that a competent lawyer was expected to understand. AI has now joined that list under Prof.Cond.R. 1.1 and Comment [8].
That framing matters. It means the duty of competence requires you to keep abreast of the benefits and risks of AI, which the Guide says you satisfy through continuing education, consultation with proficient lawyers and experts, and a degree of self-study. At a minimum, you should be able to explain how your chosen tool functions, its potential to hallucinate, and what datasets it trains on.
The Guide is blunt about the floor. A lawyer should never rely solely on the output of an AI tool without independently verifying it against the original source materials. Human review of facts, citations, and propositions of law is not optional.
Confidentiality is where free tools get you in trouble
This is the section most likely to cause heartburn, and it should. Under Prof.Cond.R. 1.6, you must safeguard client information against unauthorized access and inadvertent disclosure. The Guide explains that many AI tools, particularly free or public versions, treat your prompts as training data and, in some cases, as the vendor's property under their terms of service.
The practical guidance is concrete. Read the terms of service, privacy policy, and security practices before you input anything. Use only tools that explicitly state they do not store, train on, or share your prompts and outputs, with the Guide pointing to enterprise legal products as examples. Anonymize client information as a default, and in some matters obtain informed consent before inputting client data at all.
There is a subtle trap the Guide flags that catches even careful lawyers. Even a hypothetical, stripped of names, can inadvertently reveal enough to identify a client when the tool surfaces it in response to another user's prompt. If you would not write it in a published column, do not type it into a public chatbot.
Telling the client, and billing the client
On disclosure, the Guide takes a measured position. Using AI for legal research does not, by itself, require telling the client. Disclosure becomes more likely when the use materially affects how the representation is conducted, when client information will be shared through a tool, or when the client will be separately billed for AI expenses. And you must always answer honestly when a client asks whether you are using AI, per Prof.Cond.R. 1.4(a)(4). If a client asks you not to use AI in their matter, respect it.
On fees, the Guide draws a line that flat fee practitioners in particular need to internalize. You may bill only for time actually spent, not the time a task would have taken without AI. The example is unambiguous. If a will and trust that once took 15 to 20 hours now takes 8, you bill 8. The efficiency gain belongs to the client. The Guide also notes that as adoption grows, the profession may shift further toward flat and alternative fees, but those fees must still be reasonable. If a tool dramatically reduces the effort a task requires, charging the old flat fee may no longer be defensible.
The cost of a standard AI subscription is generally overhead you cannot pass through. The narrow exception is sophisticated per-use tools, like predictive case or jury analysis, where a reasonable charge can be billed if disclosed in the engagement letter consistent with Prof.Cond.R. 1.5.
The hallucination problem has reached Ohio
The most important development in this Guide is local. For two years, the cautionary tales came from elsewhere — Mata v. Avianca in the Southern District of New York, People v. Crabill in Colorado. Ohio now has its own.
In State v. Coleman, 2026-Ohio-965, a lawyer was sanctioned $2,000 and referred to the Office of Disciplinary Counsel for submitting filings with hallucinated facts and quotations that did not appear in the record. (Single-source citation as presented in the Guide; verify against the official Ohio reporter and a second source before relying on it in any filing.)
The Guide ties this conduct to Prof.Cond.R. 3.1, 3.3, 8.4(c), and 8.4(d). A filing built on a fabricated citation is a false representation of law to the tribunal and, because it has no basis in law or fact, is frivolous. The duties of competence and diligence under Prof.Cond.R. 1.1 and 1.3 require you to cross-check every citation, quote, holding, and proposition of law an AI tool produces. The same verification applies when you use AI to summarize pleadings, briefs, or transcripts. And if you discover a hallucination already in a filing, you must promptly notify the court and opposing counsel. Silence compounds a mistake into misconduct.
The Guide's bottom line on research is one practitioners may resist but should hear clearly. As of publication, general and publicly accessible AI is not a reliable substitute for traditional legal research, and using it for that purpose is not a best use case given the current state of the technology. The Guide points instead toward commercial databases like Westlaw and Lexis that build on defined datasets of published decisions, current statutes, and reliable secondary sources — and even those results require verification.
Finally, watch your local rules. The Guide cites a Cuyahoga County common pleas certification requirement and reminds lawyers under Prof.Cond.R. 3.4 to track standing orders on AI use and disclosure in every court where they appear.
Supervision does not stop at your firm's door
Under Prof.Cond.R. 5.1 and 5.3, managing and supervising lawyers must make reasonable efforts to ensure that lawyers and nonlawyer staff comply with the Rules. The Guide extends this to AI directly. The fact that a tool is operated by a third-party vendor does not relieve you of responsibility for its output. You should ask whether a vendor's tool can satisfy your ethical obligations to the same extent as a human you would hire for the same task, and you must ensure vendors are bound by confidentiality obligations consistent with Prof.Cond.R. 1.6 and 5.3.
Competent AI use is a firm-level discipline. It requires written internal policy, training, and ongoing supervision, not a vague trust that everyone will be careful.
This is the structural point that gets lost in conversations about individual prompts. Competent AI use is a firm-level discipline. It requires written internal policy, training, and ongoing supervision, not a vague trust that everyone will be careful.
Judges get their own warning
Roughly half the Guide addresses the bench, and the message is more restrictive. A judge must perform judicial duties competently under Jud.Cond.R. 2.5(A), which now includes understanding AI's capabilities, limitations, and risks. But the core caution is about decision-making itself. Under Jud.Cond.R. 2.7, a judge has a duty to sit and to decide, and the Guide concludes a judge should never use AI to reach the ultimate decision in a case or to produce the first draft of a decision or order. Doing so risks predisposing the judge before independent analysis and denies litigants their right to be heard by the judge rather than by a machine.
The Guide raises distinct judicial concerns that lawyers should understand, because they shape how courts will view AI generally. AI can carry human bias into a result. General tools can expose a judge to ex parte information outside the record, implicating Jud.Cond.R. 2.9 and potential disqualification under 2.11. And AI's tendency to build an argument around the conclusion implied by a prompt, useful to an advocate, is corrosive to neutral judicial reasoning. The Guide's memorable instruction is to treat AI like a first-year law clerk and double-check everything.
What to do this quarter
The Guide does not require you to abandon these tools. It requires you to govern them. Concretely, that means a short list every Ohio firm can act on now.
Audit your tools
Read every terms of service and privacy policy for anything that touches client information. Confirm it does not train on or share your prompts.
Adopt a written AI policy
Cover permitted tools, confidentiality protocols, verification requirements, and supervision duties under Prof.Cond.R. 5.1 and 5.3.
Build citation verification into workflow
No AI-assisted filing leaves the office without a human confirming every authority against an original source.
Revisit flat fees and engagement language
Fees must reflect actual time spent. Any per-use tool costs must be disclosed in the engagement letter consistent with Prof.Cond.R. 1.5.
Calendar local rule checks
Check standing orders and local rules on AI use and disclosure in every court where you practice, and set recurring reminders.
The lawyers who get into trouble with AI are not the ones who used it. They are the ones who used it without verifying. Ohio has now said as much in plain language. The standard of care has moved, and the Ethics Guide is the map for staying on the right side of it.
The Bottom Line
Ohio's Ethics Guide is the most practical, jurisdiction-specific guidance on AI and legal ethics published in the United States to date. It does not ban AI. It disciplines it. Treat it as your firm's baseline operating manual for AI governance — because the next State v. Coleman may carry your name on the caption.
Disclaimer: This article is for general informational purposes and does not constitute legal advice. The Ohio Board of Professional Conduct's Ethics Guide is nonbinding staff guidance and does not reflect the views of the Board, its Commissioners, or the Supreme Court of Ohio. Lawyers should consult the Guide and the applicable Rules directly and verify all citations against original sources before relying on them. LegalTek.ai is a technology company, not a law firm. No attorney-client relationship is created by reading this material.
Appendix: Sources
- Ohio Board of Professional Conduct, Artificial Intelligence for Lawyers and Judicial Officers (Ohio Ethics Guide, published April 2026), https://www.bpc.ohio.gov/_files/ugd/c6a571_4e12e5f9a0104eaebe9351007c899b4f.pdf
- Supreme Court of Ohio, Artificial Intelligence Resource Library
- Cuyahoga County Court of Common Pleas, Certificate re Generative Artificial Intelligence, https://cp.cuyahogacounty.gov/media/3728/jjr-certificate-re-generative-artificial-intelligence.pdf