Clio’s Judiciary AI Push Hits a 50-State Split

Clio wants to sell AI to judges. State court systems do not agree on whether judges may use it. That tension is the news, not a verdict on any product.

On November 5, LawSites published Bob Ambrogi’s 50-state survey of generative AI policies aimed at judges, magistrates, law clerks, court administrators and similar court personnel. The piece sits next to Clio’s judiciary build-out: senior hires, a trademarked phrase “judiciary-grade AI,” and the September acquisition of Learned Hand, which Legal Tech Circle covered earlier. The survey’s point is simpler. Courthouse doors are not equally open, and in at least one state they are closed for legal work.

What the LawSites Survey Found

Ambrogi’s survey, current as of October 30, 2026 by his account, sorts states into binding policies, ethics guidance, administrative principles, developing efforts, or no publicly identified statewide instrument. He used AI tools to help compile the chart, then checked results with another model and with his own research, and invited corrections. “Not identified” does not prove a policy is absent. It means none was found in public sources.

The spectrum is wide. Massachusetts is described as the most restrictive: generative AI may not be used for legal research, legal writing or other legal work by covered court personnel. New York is described as more permissive, with approved tools under conditions and a September 2026 ethics opinion that recognizes using private AI to summarize judicial proceeding transcripts. Rhode Island is cited for detailed but expressly advisory judicial ethics guidelines. Several states, including Arizona, New York, Pennsylvania, South Carolina, Utah and Virginia in Ambrogi’s write-up, have controls over which tools court personnel may use. Training, supervisory permission or disclosure of staff AI use to the assigning judge appear in more than one policy family.

One point of overlap across the policies Ambrogi summarizes is that judges may not delegate judgment to AI. Assistance with research, summaries or drafting, where allowed, still leaves the judge responsible for the record, the law and the decision. Verification of citations and facts shows up repeatedly, as does a common ban on putting nonpublic case information, sealed material or personally identifying information into a public chatbot.

Ambrogi’s practical question for vendors is whether judges may use AI to analyze evidence, summarize testimony, research law, draft opinions or assist with factual findings. Scheduling notes and routine correspondence are the easy cases. Substantive judicial work is where the state rules diverge.

Massachusetts Sets the Hard Line

The Massachusetts rule is not a blog summary. The Supreme Judicial Court’s Interim Guidelines for use of Generative AI, dated November 12, 2025, apply to judges, clerks, registers, recorders, employees, law clerks, interns and contractors of the Massachusetts state courts. On legal research, legal writing or other legal work, the answer is “Not at this time.”

Limited administrative use is allowed only with approved tools, privacy settings that opt out of training and retention, and Level 1 public data. The listed tools on the public guidance page are OpenAI’s ChatGPT, Google Gemini and Microsoft CoPilot. Outputs must be independently verified. Users remain responsible under ethical codes and court policies. The document says the guidelines are interim and “will be refined and modified as more becomes known about Generative AI.” Ambrogi wrote that a check with the court the week before his November 5 article found no modifications.

Massachusetts Lawyers Weekly covered the same guidelines in January 2026, reporting that GenAI is barred for legal research or writing and limited to administrative tasks with public data in approved tools. That reporting matches the primary text. For lawyers and vendors, the operational effect is clear: a court employee in Massachusetts cannot lawfully use GenAI for the core legal tasks many judiciary AI products advertise, at least while these interim guidelines stand.

What This Means for Clio and for Counsel

Clio’s judiciary push does not change any state’s rules. Learned Hand and whatever Clio builds next still have to fit local court policy, procurement and ethics guidance. In Massachusetts, a product aimed at judicial legal research or opinion drafting faces a categorical interim bar for covered personnel. In New York or other approved-tool states, the path is conditional: authorization, training, confidentiality limits and human verification. In states with no identified statewide policy, local courts and individual chambers may still set their own limits, and silence is not permission.

For litigators, the survey is a map of where opposing judges and clerks may or may not use assistance tools. It does not tell you which chambers have adopted which vendor. It does tell you that a national “judiciary AI” sales pitch will meet fifty different compliance conversations, and that verification duties remain even where use is allowed. For court administrators evaluating vendors, the Massachusetts text is a useful stress test: if a tool’s main value is legal research or legal writing for judges, that value is blocked today in at least one large state market.

Legal Tech Circle is not scoring Clio’s strategy here. The facts on the table are a vendor moving into courts and a patchwork of judicial GenAI rules that already constrain how far any tool can travel without local approval.

Common Questions

Does any state ban judges from using GenAI for legal work?

Yes. Massachusetts interim guidelines say GenAI may not be used for legal research, legal writing or other legal work by covered court personnel. Limited administrative use with public data and approved tools is still allowed.

Does the LawSites survey cover lawyer or litigant AI rules?

No. Ambrogi limited the survey to policies directed at judges and judicial personnel, not attorney or filing rules aimed only at parties.

Has Massachusetts updated the November 2025 guidelines?

Ambrogi reported that a court check the week before his November 5, 2026 article found no modifications. The public Mass.gov page still describes the guidelines as interim.

Does Clio’s Learned Hand deal override state court AI rules?

No. An acquisition does not change judicial branch policy. Products still have to comply with each court’s rules, approved-tool lists and ethics guidance.

Verdict

Real. Clio is building a judiciary business, and state courts already disagree on whether judges may use generative AI for legal work. The LawSites survey and the Massachusetts primary guidelines are concrete. What remains open is how fast restrictive states revise interim bans, which approved-tool regimes become the practical market path, and how vendors document compliance when the same product faces both a ban and a green light across a state line.

*Updated Nov 6 to correct the Legal Tech Circle Learned Hand story link.*