Docket Summary

• Senate Judiciary Chairman Chuck Grassley demands answers from two federal judges after error-riddled court orders suggest unchecked AI use—reversing the recent trend of attorneys being sanctioned for AI hallucinations
• U.S. District Judge Henry T. Wingate allegedly backdated a corrected order and scrubbed the original from public records after Mississippi's Attorney General identified fake parties, misquoted laws, and phantom individuals in a DEI ruling
• The controversy erupts as California's September 2025 AI rules for judges take effect, raising urgent questions about whether judicial delegation to algorithms threatens constitutional due process

WASHINGTON - For two years, federal courts have sanctioned attorneys who submitted briefs contaminated with AI hallucinations—fake citations, nonexistent cases, fabricated quotes. Now the tables have turned. Two Article III judges face Congressional scrutiny for allegedly using generative AI to draft substantive court orders without adequate human review, producing errors so egregious they triggered cover-up allegations and Constitutional concerns.

On October 3, 2025, Senator Chuck Grassley launched a formal oversight inquiry into U.S. District Judges Henry T. Wingate (Southern District of Mississippi) and Julien Xavier Neals (District of New Jersey), demanding explanations for court orders riddled with factual inaccuracies that legal experts say bear the hallmarks of unvetted AI output.

"No less than the attorneys who appear before them, judges must be held to the highest standards of integrity, candor, and factual accuracy," Grassley wrote in separate letters to both judges. "Indeed, Article III judges should be held to a higher standard, given the binding force of their rulings on the rights and obligations of litigants before them."

Read Grassley’s letter to Wingate HERE and letter to Neals HERE.

The more serious case involves Judge Wingate's July 20, 2025 temporary restraining order blocking Mississippi's ban on diversity, equity and inclusion programs in public schools. Mississippi Attorney General Lynn Fitch's office filed a motion documenting four categories of errors that suggest AI fabrication:

Phantom parties: The order named plaintiffs and defendants who were never involved in the case.

Misquoted statutes: State law provisions were inaccurately cited.

Unsupported factual claims: The ruling included assertions with no basis in the evidentiary record.

Ghost individuals: Four people referenced in the order do not appear anywhere in the case file.

When confronted with these errors, Judge Wingate's response raised additional red flags. He issued a "corrected" version of the order—but backdated it to the original July 20 filing date. More troubling: Wingate removed the error-filled original from the public docket entirely, erasing the transparent record of the court's mistake. His only explanation dismissed the fabrications as "clerical" errors, with no acknowledgment of how such systematic inaccuracies could occur.

Legal ethics experts note that backdating court documents and scrubbing public records suggests consciousness of serious misconduct. "This isn't a typo," said one federal practice attorney who requested anonymity. "You don't accidentally invent four human beings and multiple parties to a lawsuit. And you definitely don't hide the evidence when you get caught."

Judge Neals' case in a biopharma securities matter followed a different trajectory—but reached the same disturbing destination. On July 23, 2025, Neals withdrew his entire decision after defense counsel identified multiple AI-characteristic errors:

Fabricated quotes: The opinion attributed statements to defendants they never made.

Phantom precedent: The ruling cited quotes from judicial decisions that don't actually contain those passages.

Reversed outcomes: The opinion misstated case results, claiming motions to dismiss were denied when court records show they were granted.

Unlike Wingate, sources close to the New Jersey court offered an explanation—one that may be worse than silence. According to reporting, "a person familiar with the matter" revealed that "a temporary assistant" in Neals' chambers "used an artificial intelligence platform" to draft the decision, and "the opinion was inadvertently issued before a review process was able to catch errors introduced by AI."

This admission raises immediate questions: What review process existed if the opinion was issued before it could catch obvious errors? And most fundamentally: Is a federal judge's constitutional obligation to decide cases delegable to algorithms supervised by temporary staff?

California's AI Rules: Model or Cautionary Tale?

The timing of these scandals is particularly significant. On September 1, 2025, California became the first state to implement comprehensive rules governing judicial use of AI, adopting California Rules of Court, rule 10.430 and California Standards of Judicial Administration, standard 10.80 based on recommendations from the state's Artificial Intelligence Task Force.

Proponents tout these rules as necessary guardrails for inevitable technological integration, and a pathway for Judges and their staff to use AI in their judicial work. Critics see something darker: official approval for judges to offload their constitutional duties onto algorithms, with predictable consequences for accuracy and accountability.

The California framework arrives as federal courts grapple with the Wingate and Neals scandals—cases that expose the risks when judicial AI use lacks transparency, oversight, and human verification. The question facing other states considering similar rules: Are they preventing the next AI disaster, or codifying the conditions that make it inevitable?

The Constitutional Stakes

Federal judges wield extraordinary power. A single order can strip constitutional rights, freeze assets, separate families, or jail defendants. Unlike attorneys—who can be sanctioned, sued for malpractice, or disbarred—Article III judges enjoy life tenure and removal only through impeachment.

That power demands corresponding responsibility. When attorneys submitted AI-generated briefs containing fake cases, judges sanctioned them, and in some case referred them to bar associations. But when judges themselves use AI without adequate oversight, who holds them accountable?

Grassley's letters demand answers to specific questions:

• Did the judges, their law clerks, or court staff use generative AI in preparing the disputed orders?
• Was non-public case information entered into AI platforms, potentially violating confidentiality and privacy rules?
• Will the courts re-docket the original error-filled orders to preserve a transparent record?

The Senate Judiciary Committee's jurisdiction over federal courts and judicial proceedings gives Grassley leverage—but not direct disciplinary authority. That power rests with the Judicial Conference and, ultimately, with impeachment proceedings that require substantial political will.

The Accountability Gap

The Wingate case is particularly troubling because it suggests active concealment. Backdating a corrected order and removing the original from public access doesn't just hide mistakes—it obscures evidence of how those mistakes occurred. If judges can sanitize the docket after AI-generated errors come to light, the public and litigants lose any ability to verify the integrity of judicial decision-making.

This creates an accountability asymmetry. When lawyers use AI recklessly, the evidence remains in the court record, exposing them to sanctions. When judges do the same, they control the docket and can potentially erase the proof.

The New Jersey case offers a different lesson: delegation without supervision. If the explanation provided to media is accurate—that a temporary assistant used AI to draft an opinion later issued without adequate review—it reveals a systemic breakdown in judicial chambers. Federal judges don't draft every word themselves; they rely on law clerks and staff. But the ultimate responsibility for accuracy and legal reasoning cannot be delegated, especially not to algorithms.

What's Next

Senator Grassley has requested responses from both judges, though no deadline is specified in public materials. The Senate Judiciary Committee lacks direct authority to discipline federal judges but can:

• Refer findings to the Judicial Conference for ethics investigation
• Hold public hearings highlighting systemic problems with judicial AI use
• Condition future judicial funding on transparency and oversight requirements
• Recommend impeachment proceedings to the House if evidence supports it

The broader question remains unanswered: Can judicial AI use be regulated effectively, or does the technology's propensity for hallucination make it fundamentally incompatible with the judicial function?