Inside N.D. Ohio's Total AI Ban: Judge Boyko's Standing Order and What It Actually Prohibits

Written by the Citation Safe Research Desk · Reviewed by Andy Gaber, Founder — July 17, 2026

Part of: Court AI Standing Orders by Jurisdiction

Judge Christopher A. Boyko of the Northern District of Ohio has one of the strictest AI standing orders in the federal system: it bans AI outright from “the preparation of any filing submitted to the Court,” rather than requiring disclosure or certification.
The order carves out ordinary legal and internet search engines — Westlaw, LexisNexis, Google, Bing — from the ban, targeting generative drafting and research tools specifically, not any tool that happens to use AI somewhere in its infrastructure.
Violations can draw sanctions “including, inter alia, striking the pleading from the record, the imposition of economic sanctions or contempt, and dismissal of the lawsuit” — a broader and more severe consequence menu than most certification-model orders specify.
The Ropes & Gray tracker’s count of only 5 entries prohibiting AI use outright, against 756 that address AI use in some fashion, confirms the Boyko model remains a distinct minority approach even as the broader body of AI-related court rules has grown.

What “any use” actually covers

The Boyko order’s operative language is broad by design: “no attorney for a party, or a pro se party, may use Artificial Intelligence (’AI’) in the preparation of any filing submitted to the Court” (full text at ohnd.uscourts.gov). That phrasing — “preparation of any filing” — reaches further than a rule limited to drafting the final document text. Preparation plausibly covers research assistance, outlining, summarization of case law, and drafting, not merely the act of generating the final prose that appears on the page. A certification-model order (like Judge Starr’s or Judge Wang’s) asks attorneys to disclose and verify AI use; the Boyko order removes the option to use AI in filing preparation at all, regardless of whether the output would later be verified.

The carve-out that keeps the order from banning legal research itself

A rule this broad would be nearly unworkable if it swept in every tool an attorney might touch during filing preparation, since AI features have become embedded in mainstream legal research platforms. The order addresses this directly with an explicit carve-out for legal and internet search engines — Westlaw, LexisNexis, Google, and Bing are named specifically as exempted from the ban (ohnd.uscourts.gov, Boyko Standing Order). The distinction the order draws is functional, not brand-based: it targets generative drafting and research tools — tools that produce new synthesized text, summaries, or citations in response to a prompt — while leaving conventional search and retrieval tools untouched, even though some of those platforms (Westlaw and LexisNexis both now ship generative-AI-assisted research features) sit right at the boundary the order is trying to draw.

That boundary is worth flagging as a genuine interpretive question rather than a settled one: if an attorney uses Westlaw’s own AI-assisted research feature — the same feature the Stanford RegLab/HAI study found produced hallucinated content in roughly a third of tested queries — does that fall inside the named carve-out (“Westlaw”) or outside it (because it’s a generative AI feature, not conventional search)? The order’s plain text names the platform, not a specific feature within it, which is a real ambiguity for any attorney trying to comply in good faith before this district’s own case law resolves the question, if it ever does.

The sanctions the order authorizes

Violations of the Boyko order can draw sanctions “including, inter alia, striking the pleading from the record, the imposition of economic sanctions or contempt, and dismissal of the lawsuit” (ohnd.uscourts.gov). Dismissal of the lawsuit is the most severe item on that list, and its inclusion sets this order apart from most certification-model orders, which typically limit their stated consequence to striking the non-compliant filing (with leave to re-file, in Judge Wang’s order, for instance). An outright ban paired with dismissal as an available sanction reflects a fundamentally different risk posture: rather than managing AI use through disclosure and accuracy verification, the order treats any use at all as the sanctionable event, with the case itself as potential collateral.

Why this remains a minority approach

The pillar’s review of the Ropes & Gray tracker found only 5 entries tagged “Prohibits Use of AI” outright, against 756 entries that apply to AI use in some fashion and 161 requiring disclosure or verification specifically (Ropes & Gray AI Court Order Tracker). That imbalance suggests most courts that have engaged with the issue have concluded a certification or disclosure model addresses the underlying risk (fabricated or inaccurate content reaching a filing) without the practical cost of a categorical ban — which, given how embedded AI-assisted features have become in mainstream legal research tools, is a harder line to draw and enforce cleanly, as the Westlaw/LexisNexis carve-out ambiguity above illustrates.

What this means for anyone filing in N.D. Ohio, or before a similarly-structured order

Confirm whether a specific judge has adopted a total-ban order before assuming a certification model applies. The Boyko order is a single judge’s standing order, not a district-wide rule — check the assigned judge’s own chambers rules, not just the district’s general local rules.
The Westlaw/LexisNexis carve-out may not extend to those platforms’ AI-specific features. Treat any AI-assisted research feature within an otherwise-exempted platform as a genuine gray area rather than a clearly permitted use, absent controlling case law resolving the question.
Dismissal is an available sanction, not a remote hypothetical. The order’s own text lists it explicitly; this is a materially higher-stakes violation than most certification-model non-compliance, where striking-with-leave-to-refile is the more common consequence.
“Preparation of any filing” is broader than “drafting the final text.” Research assistance and outlining plausibly fall inside the ban’s scope even if an attorney writes the final prose entirely independently.

Sources

Judge Christopher A. Boyko, Court’s Standing Order on the Use of Generative AI, N.D. Ohio — full text fetched and quoted directly.
Ropes & Gray, AI Court Order Tracker, category filter counts accessed 2026-07-16.

Citation Safe is a verification workflow tool, not legal advice. No attorney-client relationship is created. Human review is always required. Whether a specific AI-assisted feature within an exempted platform (e.g., Westlaw’s AI-Assisted Research) falls inside or outside this order’s carve-out is presented here as an open interpretive question, not a resolved one; confirm current guidance with the court before relying on any interpretation.

Check a brief before you file it →