Judge Richard Stearns needed four pages to hold that a university may violate Title VI for two academic years, retain every federal dollar it received, and defeat the United States by merely behaving itself once the government’s letter arrives. His order dismissing United States v. Harvard is brief, and that brevity is load-bearing: the reasoning cannot withstand a longer look at the statute, precedent, or his own prior findings.
In August 2024, in Kestenbaum v. President and Fellows of Harvard College, the same judge reviewed the same campus and refused to dismiss a private Title VI suit brought by Jewish students. He called Harvard’s response to their harassment “at best, indecisive, vacillating, and at times internally contradictory.” He warned that crediting the university’s public statements would “reward Harvard for virtuous public declarations that for the most part… proved hollow when it came to taking disciplinary measures against offending students and faculty.” His bottom line traveled around the world: “the facts as pled show that Harvard failed its Jewish students.”
Those facts barely rate a footnote, which declines to “further rehearse” the 2023-24 allegations because prior opinions, including his own, had already fleshed them out. What matters now, the court says, is the calendar. The government alleged only three later incidents, all in March 2025, and none after June 30, 2025, when the agency formally advised Harvard of its failure to comply. Those incidents were “too isolated and episodic” to show continuing institutional noncompliance; without an ongoing violation, there was no case.
The quiet the court found so probative arrived under guard. By March 2025, Harvard was operating under two antisemitism settlements, an Office for Civil Rights investigation, a presidential task force, and undivided federal scrutiny. The Kestenbaum record had already described Jewish students removing visible markers of Jewish identity and avoiding Jewish events. Incident counts fall when the targets hide. Harvard’s own amended complaint in its funding lawsuit admitted that there is still “much work to do.”
Then came a public warning against mistaking silence for compliance. On July 2, 2026, six weeks before this order, 170 Harvard faculty and staff published an open letter reporting that during the previous year, the very period the court reads as recovery, Jews and Israelis wore baseball caps over their kippot, hid their Stars of David, and scrubbed Jewish-sounding names or activities from their resumes. Students describing these experiences insisted on anonymity for fear of retribution. The essay published alongside the letter supplied the headline: “Harvard’s Antisemitism Problem Has Gone Underground.” Its point was simple: when hate goes quiet, it has either disappeared or gone underground. Harvard’s own community reported the latter.
The closest doctrinal analog points the other way. Under the Supreme Court’s voluntary cessation cases, a defendant claiming reform bears a “formidable” burden to show that the conduct will not recur. The doctrine does not govern a motion to dismiss, but it names the danger: treating temporary quiet as conclusive proof of reform. This order drew that inference from silence, then asked the government to disprove it at the pleading stage.
Here is what the order quotes from 42 U.S.C. § 2000d-1: enforcement may proceed only after the agency “has advised the appropriate person or persons of the failure to comply with the requirement and has determined that compliance cannot be secured by voluntary means.” Here is what it does not quote: any language requiring the failure to continue after notice. Why? Because no such language exists. The statute speaks of “the failure to comply,” the completed event the notice addresses. The ongoing-violation requirement that decided this case appears nowhere in the text Congress wrote. The order imported a mootness-like requirement into a notice provision.
The recoupment analysis is even more vulnerable. The order assumes, without deciding, that the government may use “any other means authorized by law” to recover funds Harvard drew during the violation years, then rejects that backward-looking claim because nothing happened after notice. But recoupment is retrospective by definition, and notice necessarily postdates the violations it identifies. Bell v. New Jersey and Bennett v. Kentucky Department of Education held in the federal-grant context that recovering funds obtained on broken assurances collects a debt, not a penalty. Barnes v. Gorman describes Title VI as Spending Clause legislation “much in the nature of a contract” and says recipients are on notice of remedies traditionally available for breach. Together, those cases point sharply against the order’s premise. At a minimum, they demanded analysis. Instead, the order makes recoupment effectively unavailable whenever a recipient cures before notice. Getting caught now cures everything.
Footnote three deserves its own appeal. This was a motion to dismiss, where well-pleaded facts are accepted as true and reasonable inferences are drawn for the government. Yet the court faulted the government for failing to identify “any rational basis” for the agency’s determination that voluntary compliance could not be secured. APA reasonableness review ordinarily proceeds on an administrative record, often at summary judgment. When Harvard challenged the funding freeze in President and Fellows of Harvard College v. HHS, it received full APA review on cross-motions for summary judgment and won. Here, the government’s determination was dispatched in a footnote, without a record and with the inference drawn against the pleader. Footnote three gets one thing right: refusing the government’s demands is not, by itself, noncompliance with Title VI. But the complaint alleged the statutory notice and the statutory determination, and a motion to dismiss takes those allegations as true. The footnote did not find them implausible. It demanded the proof behind them, before any rule required proof at all.
The government warned that the ruling would make it “virtually impossible for the United States to win a Title VI suit because a reasonable defendant would cease its indifference as soon as the United States threatens to withhold funding.” The court called that a “fevered forewarning of potential havoc,” then adopted the forecast and declared it a happy ending: a defendant scared into behaving is the statute working as intended. The order insists it is “limited to the specific circumstances presented here.”
But violations, followed by notice, followed by temporary quiet, will recur whenever a sophisticated recipient changes course when funding is threatened. And if the problem was only the age of the allegations, it could have been cured by amendment. Yet the order dismissed the complaint outright without addressing futility or granting leave to amend. A pleading defect became a substantive amnesty.
The government should appeal. Rule 12(b)(6) dismissals are reviewed de novo, and the statutory text and Supreme Court precedent will be waiting at the First Circuit. It should also investigate the post-notice incidents identified in the faculty letter and, if the facts support it, bring a new action. Finally, the agency may pursue § 2000d-1’s administrative termination process, with a hearing, an on-the-record finding, and a report to Congress, subject to judicial review on a fully developed record. The government should preserve all three paths.
The judge who found in 2024 that Harvard failed its Jewish students has now ruled that the United States arrived too late to impose any consequences for it. The court of appeals should explain the difference between a deadline Congress enacted and one that was invented.
***
Mark Goldfeder is CEO of the National Jewish Advocacy Center and a law professor at Touro Law School. @markgoldfeder on X
