DOJ's Subpoena Errors Show a Pattern of Sanctionable Conduct

The useful moment in the New York Times subpoena hearing was not the government’s retreat by itself. It was Judge Arun Subramanian reportedly telling the Justice Department that, “if this were a civil proceeding,” he would ask the parties to show cause why sanctions should not issue for “incorrect statements or misstatements” in court filings, and then asking whether the department would “do this again.” Assistant U.S. Attorney Kevin Sullivan answered, “No, your honor.” [1][2]

That is the right place to start because it keeps the issue where lawyers actually have to manage it: not at the level of press-freedom abstraction, but at the level of filings, certifications, factual support, sealed applications, date ranges, and process choices. The hearing details discussed here come from same-day reporting by PBS, Courthouse News, NBC News, NPR, and CBS News. As of July 27, 2026, a written order was not available in the materials reviewed, so the judge’s comments should be treated as reported hearing statements, not as the text of an entered sanctions order. [1][2][3][4][5]

There is another limit worth putting on the page early: Federal Rule of Civil Procedure 11 does not directly govern criminal grand-jury practice. Judge Subramanian’s civil-proceeding comparison is still important because federal litigators know exactly what that comparison means. When a judge says the same conduct would invite a sanctions show-cause order in civil court, the practical signal is not decorative. It is a warning that the government’s paper, if tested under ordinary court-facing standards, would have serious evidentiary-support and candor problems.

Six errors, one accumulating risk problem

The DOJ withdrew the subpoenas after the judge pressed the government on a cluster of defects. Each error could matter on its own. Together, they make the episode look less like a single bad draft and more like compulsory process issued without the file discipline that the governing rule itself requires.

The non-disclosure application is the hardest problem to minimize

The government’s most exposed position appears to be the sealed non-disclosure application. PBS and Courthouse News reported that the application said the investigation was not public, even though multiple national outlets had already reported on it. Sullivan, who had to answer questions from the gallery, called the statement an oversight. [1][2]

That explanation may matter to intent. It does not solve the court-facing problem. A lawyer seeking a non-disclosure order is asking a judge to restrict what a subpoena recipient can say and when. In that posture, the assertion that an investigation is not public is not filler. It is part of the justification for secrecy. If the assertion is wrong, the immediate question is not whether the lawyer was embarrassed. It is whether the statement was reasonably checked before it was filed.

This is why Judge Subramanian’s sanctions language landed with force. Rule 11(b)(3)-style analysis asks whether factual contentions have evidentiary support after reasonable inquiry. Inherent-authority analysis asks different questions, including whether conduct reflects bad faith or an abuse of the judicial process. The hearing reports do not establish that sanctions were entered, or that the judge made a finding of bad faith. They do show why a judge would view an incorrect sealed filing as more than a clerical inconvenience.

Two days is a thin record for “last resort” process

The timing problem is simple enough to be damaging. The first New York Times article published on July 8, 2026. The subpoenas went out July 10. The regulation governing compulsory process for news-media records requires that “all reasonable attempts” be made to obtain the information from alternative sources before seeking information from members of the news media. [1][6]

A subpoena served two days after publication may be defensible in some emergency record, but that defense has to be built. The file should be able to show what alternative sources were considered, who made the decision that they were inadequate, and why further pursuit would not work. Without that paper trail, “last resort” becomes a label attached after the fact.

For counsel challenging similar process, this is often the first pressure point. Do not only ask whether the government cited the regulation. Ask what the government did before it decided the reporter’s records were necessary. A court evaluating compulsory process is entitled to more than a conclusory assurance that alternatives were exhausted.

Negotiation rules require a record, not an attitude

The pre-subpoena negotiation defect is closely related but not identical. Section 50.10 requires negotiations with the affected member of the news media unless an exception applies, including a national-security exception that must be certified. The hearing reports state that the government did not produce such a certification. [1][6]

That is a familiar litigation-risk failure. If a rule says a lawyer may skip a procedural protection only after a specified showing, the protection is not optional just because the subject matter is sensitive. The certification is the mechanism that lets the court and the recipient understand which path the government claims to be using.

This also matters because negotiation is not merely courtesy. It can narrow custodians, reduce date ranges, clarify whether the recipient has the records, and avoid service on people who should never have been swept in. When the government skips negotiation and then serves overbroad process, the missed conversation becomes part of the causation chain.

Family subpoenas turn weak matching into personal burden

The subpoenas reportedly reached beyond the reporters to spouses and one reporter’s mother, based on uncorroborated public-source database matches. Courthouse News reported Judge Subramanian’s concern that “inadvertent errors” were “piling up.” [1][2]

There is a practical reason this error reads differently from an overlong date range. A family-member subpoena imposes costs on people who may have no role in the reporting, no professional obligation to preserve source relationships, and no reason to expect that a leak investigation will arrive at the front door. The government’s evidentiary threshold should rise before compulsory process moves from a reporter to a spouse or parent.

Public-source database matches can be useful leads. The problem is treating a lead as if it were a verified basis for process. In ordinary federal practice, that is where a Rule 11(b)(3)-style inquiry becomes concrete: who checked the match, what corroborated it, what contrary information was available, and why the filing or subpoena did not disclose the uncertainty.

The January 1 range and the home service add context

The subpoenas sought records beginning January 1, 2026, even though the articles at issue were published in July. The regulation requires that compulsory process be narrowly tailored and limited to a reasonable period. [1][6]

A January 1 start date may be defensible if the government can tie it to facts in the investigation. The problem is that broad retrospective dates require explanation. Without a record connecting the lookback period to the suspected disclosure, the date range looks like a dragnet attached to a leak theory.

The reported home service during evening family time is different. It is not, by itself, the kind of defect that carries the same doctrinal load as a false statement to a court or a missing certification. But judges are human institutional actors. When aggressive service choices sit next to overbreadth, missing negotiations, and a misleading sealed application, they affect how the whole episode reads. NBC reported that FBI agents served subpoenas at reporters’ homes, including during evening hours. [3]

The Bondi rule allowed subpoenas, but it did not erase the safeguards

The regulatory backdrop matters, but it should not swallow the analysis. The May 2, 2025 final rule, published at 90 Fed. Reg. 18786, rescinded the Biden-era prohibition on certain compulsory process directed at members of the news media in leak investigations. It did not give prosecutors a blank form. The rule preserved requirements involving last resort, alternative sources, negotiations, and narrow tailoring. [7]

That is why this episode is poorly described as a simple consequence of a policy change. The policy change reopened a door. The defects identified at the hearing concern how the government walked through it. If the operative rule permits subpoenas only after specified procedural safeguards, failure to document those safeguards makes the process vulnerable even before the court reaches broader constitutional arguments.

For the broader procedural sequence, see How the DOJ Violated Its Own Rules in the NYT Subpoena Case and DOJ Withdraws NYT Subpoenas After Judge Cites Legal Errors. For the regulatory change itself, see How the Bondi Rescission Enabled DOJ Subpoenas of Journalists.

The June withdrawals make isolation harder to argue

The New York Times subpoenas were not the only recent round that ended in withdrawal. CNN reported in June 2026 that the DOJ withdrew subpoenas targeting Washington Post reporter Ellen Nakashima and Wall Street Journal reporters after those subpoenas were challenged. CNN also reported that President Trump personally pressed acting Attorney General Todd Blanche, handing him articles marked “Treason” in Sharpie. [8]

That parallel should be handled carefully. The details in the materials reviewed rely on secondary reporting, not a full public docket record. It is enough to say that the NYT episode was at least the third reported instance in which journalist subpoenas in this broader leak-investigation setting were withdrawn after challenge. It is not enough, on this record, to assign the same six errors to each earlier subpoena.

Still, recurrence matters for risk. A one-off mistake can be corrected with a better checklist. Repeated withdrawals after challenge suggest a process that is reaching recipients before the legal review is complete. That is the kind of pattern judges notice, particularly when the government asks for trust in sealed or partially sealed proceedings.

How counsel should read the sanction signal

The immediate result was withdrawal, not a sanctions order. CBS reported that the government said the investigation remains ongoing, which leaves open the possibility that the DOJ could try again with different process. [5]

That distinction matters. This is not an entered precedent holding that Rule 11 applies to criminal grand-jury subpoenas directed at journalists. It is also not merely courtroom color. A federal judge reportedly identified incorrect or misleading filings and said comparable civil conduct would trigger a sanctions show-cause order. For lawyers who manage subpoenas, sealed applications, and non-disclosure orders, that is a usable risk record.

The practical lessons are ordinary, which is part of the point:

• Demand the exhaustion record. If the government says compulsory process was a last resort, ask what alternative sources were pursued, when, and by whom.
• Ask for the negotiation basis. If negotiations were skipped, identify the exception invoked and the certification or approval that supports it.
• Compare sealed assertions against public facts. A non-disclosure application that says an investigation is not public should be tested against public reporting before the court relies on it.
• Separate leads from verified facts. Database matches, family connections, and inferred identifiers should be challenged unless the government can show corroboration.
• Treat breadth and service choices as cumulative evidence. A long date range or evening home service may not carry the motion alone, but each can support the argument that the process was not narrowly or carefully handled.

This is also where the episode connects to sanctions practice outside leak investigations. Courts have been increasingly direct when lawyers submit unsupported factual or legal material, including in technology-driven filing failures; the same basic discipline appears in AI Citation Hallucination Sanctions in Federal Courts. The subject matter changes. The judge’s question does not: what did the lawyer do before asking the court to act?

For readers focused on the journalist-protection doctrine rather than the motion-practice failure, the related First Amendment and source-protection issues are treated in What Legally Protects Journalists From Subpoenas Like the Air Force One Case? The strategic use of labels in this investigation is discussed in The Material Witness Label Behind Trump's Journalist Subpoenas. Those issues matter, but they are not necessary to see the filing-risk problem here.

The disciplined conclusion is narrow. DOJ was not sanctioned in the materials reviewed. Judge Subramanian’s Rule 11 comparison is not a direct holding in a criminal grand-jury context. But the six-error record is concrete: speed without exhaustion, compulsory process without negotiation, family subpoenas without corroboration, broad dates without tailoring, a sealed application with an apparently false public-status assertion, and service choices that made the whole exercise look needlessly aggressive. In ordinary federal practice, that is exactly how sanction risk accumulates.

References

1. Trump administration admits legal errors and withdraws subpoenas of 3 NYT reporters — PBS
2. Ripped by judge, DOJ drops subpoenas of New York Times reporters — Courthouse News
3. Justice Department agrees to withdraw subpoenas of New York Times journalists — NBC News
4. DOJ drops subpoenas of New York Times reporters under judge's pressure — NPR
5. Government withdraws subpoenas for New York Times reporters — CBS News
6. 28 CFR § 50.10 — Cornell Legal Information Institute
7. Policy Regarding Obtaining Information From, or Records of, Members of the News Media — Federal Register, May 2, 2025
8. Trump's DOJ withdrew subpoenas targeting Washington Post and Wall Street Journal reporters — CNN