Discovery sanctions are penalties a court imposes on a party in a lawsuit for mishandling discovery — the pre-trial stage where each side must search for, preserve, and hand over evidence relevant to the case. They range from a fine to a judge simply deciding the case against the offending party, and courts use them to punish and deter cheating during fact-finding, before a jury ever hears the underlying dispute.
What counts as discovery misconduct
Discovery misconduct covers a range of behavior, not just outright destruction of evidence. Common examples include:
- Withholding a known capability. Telling the court a search or export can’t be done, when it in fact can.
- Evasive or incomplete answers. Technically responding to a request without actually answering it.
- Producing unusable evidence. Handing over documents so heavily redacted, or in such a disorganized form, that the other side can’t actually use them.
- Spoliation. Losing, altering, or destroying evidence that should have been preserved — the most serious category, sometimes called spoliation of evidence.
In US federal court, this is governed mainly by Rule 37 of the Federal Rules of Civil Procedure, with a specific 2015 amendment — Rule 37(e) — addressing the loss of electronically stored information such as emails, internal logs, or database records.
What the penalties actually look like
Courts have a menu of remedies, roughly escalating with how serious and deliberate the misconduct was:
- Monetary sanctions. The offending party pays the other side’s attorneys’ and experts’ fees spent chasing down the missing or mishandled evidence.
- Evidentiary sanctions. The judge bars the offending party from using the disputed material, or from making arguments that depend on it.
- Adverse inference instructions. The judge tells the jury it may assume the missing or hidden evidence would have hurt the offending party’s case — a significant blow, since juries tend to follow such instructions closely.
- Case-dispositive sanctions. In the most extreme cases, a judge can strike a party’s claims or defenses entirely, or enter default judgment — effectively deciding the lawsuit as a penalty, without a trial on the merits.
Under Rule 37(e), US federal courts reserve the harshest of these — the adverse inference and case-ending sanctions — for cases where a party is found to have acted with actual intent to deprive the other side of the evidence, not just carelessness.
Why this matters for AI lawsuits
Discovery has become a central battleground in the wave of copyright lawsuits against AI companies, because the evidence that matters most — what data trained a model, and what a model actually outputs — usually exists only on the AI company’s own servers. Plaintiffs can’t independently verify a company’s claims about its own logs and training pipelines; they depend on the company to search for and hand over that evidence honestly.
That dependency is exactly what makes discovery disputes so consequential in this wave of litigation. If a court finds a company misrepresented what it could search for, or handed over data in a deliberately unusable form, the resulting sanction can shape the outcome of the underlying copyright case before a jury even weighs in — for instance, through an instruction telling jurors to treat the disputed evidence as proof of copying.
In the news
A group of publishers led by The New York Times has asked a federal court to sanction OpenAI, arguing the company spent two years denying it could search its models and internal logs for the publishers’ copyrighted articles — then had an employee testify that such searches had, in fact, been run. Read our report: NYT-Led Publishers Ask Court to Sanction OpenAI Over Evidence.
Sources: Federal Rules of Civil Procedure, Rule 37 (Cornell Law School’s Legal Information Institute); Bloomberg Law reporting on In re OpenAI, Inc. Copyright Infringement Litigation.