The Tenth Circuit Is Proposing Human Review for AI-Assisted Filings

On September 18, 2026, the United States Court of Appeals for the Tenth Circuit opened public comment on proposed revisions to its 2027 local rules. One proposed change is especially important for lawyers using generative AI.

Under the proposed revision to Tenth Circuit Rule 46.5(B), a signed filing would certify that any document prepared with generative AI assistance was reviewed before filing by a human who independently verified the cited authorities, quotations, citations, legal analysis, and compliance with applicable rules and ethical duties.

That proposal is not yet the final 2027 rule. The court states that the 2027 rule changes are expected to take effect January 1, 2027, after the public-comment process and final publication.

This is not really about whether lawyers may use AI

The more useful question is what happens when AI-generated work becomes the lawyer's work.

AI does not sign the pleading. It does not place a bar number on the filing. It does not stand before the court to explain a fabricated authority, a distorted quotation, or a legal proposition that was never independently checked.

The lawyer does.

That is why I view meaningful human review less as resistance to technology and more as professional self-preservation.

AI can assist with research, organization, drafting, editing, and analysis. But the lawyer remains responsible for what is filed. The proposed rule simply makes that responsibility more explicit for AI-assisted work.

The Tenth Circuit has already been warning litigants about careless AI use

The proposal arrives against a backdrop of Tenth Circuit decisions addressing fabricated or inaccurate authorities. In a June 15, 2026 order, the court warned that litigants must read their filings and take reasonable care to avoid factual and legal misrepresentations, even when AI tools were used to prepare the submission.

The court has also emphasized that there is nothing inherently improper about using AI in legal practice. The problem is careless use that wastes judicial resources, burdens opposing parties, and damages the credibility of the legal process.

Human review protects more than the court

It protects the lawyer's credibility.

It protects the client from avoidable error.

It protects the record from fabricated authority.

It protects the lawyer's professional reputation and reduces avoidable sanction and disciplinary risk.

And it protects the very thing clients are actually paying professionals to provide: judgment.

A lawyer who simply accepts AI output has not preserved meaningful professional review. Human review requires more than reading the words on the screen. It requires checking the source, understanding the proposition, testing the analysis, recognizing what the model may have missed, and deciding whether the work should be used at all.

AI can draft. Professionals still own the work.

That principle should extend well beyond one proposed appellate rule.

Good AI governance does not require professionals to abandon powerful tools. It requires us to design the point at which automation stops and accountable professional judgment begins.

For lawyers, that is not bureaucracy. It is part of protecting the client, the court, and yourself.

Sources: Tenth Circuit public-comment notice for 2027 rule revisions and the court's full redline of the proposed 2027 rules.

This article is educational information, not legal advice. AI-assisted legal work should be reviewed in light of the applicable jurisdiction, court rules, professional obligations, and the facts of the matter.