Artificial intelligence has moved from novelty to necessity in Ohio law offices and court chambers. The question is no longer whether lawyers and judges will use it. The question is whether they will use it without violating the rules that govern their profession.
In April 2026, the Ohio Board of Professional Conduct released its Ethics Guide: Artificial Intelligence for Lawyers and Judicial Officers. The 12-page document is nonbinding, yet it carries real weight. It synthesizes sanctions orders, disciplinary cases, and emerging national consensus into practical guardrails under the Ohio Rules of Professional Conduct and the Code of Judicial Conduct. Lawyers and judges who treat the guide as optional reading do so at their peril.
Lawyers: The Duty of Competence Now Includes AI Literacy
A lawyer must provide competent representation. That duty extends to the tools the lawyer chooses. The guide makes clear that AI is now a “relevant technology” under Prof.Cond.R. 1.1, comment [8]. Competence therefore requires more than pressing “generate.” It requires understanding how the tool works, what data it trains on, and—most critically—its documented tendency to hallucinate.
Hallucinations are not rare glitches. They are the predictable output of systems that predict the next statistically likely word rather than retrieve verified law. The guide cites multiple cases in which lawyers filed briefs containing fictitious citations and propositions. One Ohio lawyer in State v. Coleman, 2026-Ohio-965, received a $2,000 sanction and a referral to disciplinary counsel after hallucinated facts and quotations appeared in court filings. The message is blunt: verify every citation, every quotation, and every proposition of law against the original source. Human review is not optional. It is the minimum.
Client Confidences and the Hidden Cost of Free Tools
Client confidentiality under Prof.Cond.R. 1.6 demands reasonable precautions against unauthorized disclosure. Many public or free AI platforms fail that test. Their terms often allow the vendor to store prompts, train future models on them, and even disclose the data. A lawyer who pastes a client’s sensitive facts—or even a carefully anonymized hypothetical—into such a tool risks turning confidential information into training data for the next user’s prompt.
The guide’s recommendation is straightforward. Use only enterprise-grade tools that explicitly promise they do not store, train on, or share prompt data. Westlaw AI, Lexis+ AI, and similar commercial products built for legal professionals meet this standard. Free ChatGPT or consumer versions do not. When in doubt, anonymize ruthlessly and consider obtaining informed client consent before any client-related information enters the system.
Independent Judgment Cannot Be Outsourced
AI can draft, summarize, and analyze at remarkable speed. It cannot exercise professional judgment. The guide warns that overreliance—especially by less experienced lawyers—turns the human into a passive reviewer rather than the author of the analysis. Judge Scott Schlegel of the Louisiana Fifth Circuit Court of Appeal puts it plainly: merely having “a human in the loop” is not enough. The quality and experience of that human oversight determines whether the output advances or undermines the client’s cause.
The same principle applies to fees. When AI cuts the time required to draft an uncomplicated will from fifteen to twenty hours to eight, the lawyer bills eight hours. The savings belong to the client. Passing on the cost of sophisticated per-use AI tools is permissible if disclosed in the engagement letter, but routine overhead is not separately billable.
Candor to the Tribunal: The Non-Negotiable Verification Rule
Prof.Cond.R. 3.3 prohibits knowingly false statements of fact or law to a court. Filing an AI-generated brief that contains hallucinated authority violates that rule and, in many jurisdictions, has already produced sanctions. Ohio courts are paying attention. At least one common pleas judge now requires counsel to certify that any generative-AI language in a filing has been checked for accuracy against print reporters or traditional legal databases by a human being.
The verification burden does not disappear when AI merely summarizes depositions or pleadings. The same cross-check against the original record is required. If a lawyer later discovers an error, immediate disclosure to the court and opposing counsel is mandatory.
Judges Face Parallel—but Distinct—Obligations
Judges must also maintain technological competence. They must understand the tools their staff and the lawyers appearing before them are using. Yet the guide draws a bright line: a judge may never delegate the ultimate decision in a case to AI. Judicial decision-making requires the exercise of independent judgment and discretion that no algorithm can replicate. Litigants are entitled to be heard by a human judge, not by a statistical model.
Additional risks are unique to the bench. AI trained on public data can surface ex parte facts never presented in the record—news articles, statistical studies, or prior commentary about the parties. A judge who uses general-purpose AI to research a pending matter risks acquiring “knowledge of disputed facts” that triggers disqualification under Jud.Cond.R. 2.11. Training data can also embed societal biases that affect impartiality. The guide therefore advises judges to treat AI like a first-year law clerk: useful for initial research or proofreading, never authoritative, and always subject to rigorous human verification.
What Forward-Looking Firms Are Doing
The most sophisticated law firms are not waiting for the next disciplinary case. They are building internal AI policies that address:
- Approved tools and prohibited tools.
- Mandatory human verification protocols (human in the loop) for all court filings and client deliverables.
- Training requirements for lawyers and nonlawyer staff.
- Clear billing guidelines that pass efficiency gains to clients.
- Vendor due diligence focused on data security and confidentiality commitments.
Firms that adopt these measures now will avoid the sanctions, reputational damage, and client complaints that will inevitably follow those that do not.
The Bottom Line
The Ohio Ethics Guide does not ban AI. It demands that lawyers and judges use it with eyes wide open. AI is a powerful assistant. It is not a substitute for legal training, professional judgment, or ethical responsibility.
Lawyers who verify every output, protect client data, and keep judgment in human hands will capture the efficiency gains without the professional risk. Judges who do the same will preserve both the reality and the appearance of impartial justice. Those who treat AI as a black box will learn its limitations the hard way—in sanctions orders, disciplinary complaints, or withdrawn opinions.
The guide is short. Read it. Implement its lessons. Then keep reading the cases and ethics opinions that follow. Technology will continue to evolve. The ethical obligations of competence, confidentiality, candor, and independent judgment will not.
Contact us
Tom Cedoz is a partner at Husch Blackwell and a member of the firm’s Artificial Intelligence Practice Group. He advises law firms and corporate legal departments on the ethical, regulatory, and practical challenges of deploying AI in legal workflows.