The brief lands in your inbox. The writing is polished, the citations are plentiful, and three of the cases don’t exist. In our latest post, we break down how to spot the tells, fix the record, and, when it matters, turn those mistakes into real leverage. From challenging credibility to positioning for fees or sanctions, it’s not about calling out the tool. It’s about holding the conduct to the standard courts already expect.
Handled well, their misuse of AI is not just a curiosity. It can be a source of leverage on the merits, on credibility, and, in some cases, on fees and sanctions. Handled poorly, it can make you look petty or technophobic. The goal is to respond in a way that advances your client’s interests and aligns you with what courts are already saying about AI, competence, and candor.
Step One: Spotting AI’s Fingerprints in Opposing Filings
Before you can exploit an AI-driven mistake, you have to recognize it. Not every weak brief is AI-generated, of course, but there are recurring tells:
- Too-perfect prose, thin substance. The writing sounds like a blog post or a generic “law school exam answer,” with broad statements of doctrine and little engagement with your actual facts or authorities.
- Citations that do not quite check out. Case names that do not appear in any database, reporters that do not exist, or real cases that say something very different than what the brief claims.
- “Frankenstein” quotes. A quotation that looks right at a glance but cannot be found in the cited opinion, or that stitches together language from multiple cases as if it came from one.
- Jurisdictional drift. Authorities from the wrong court system or era, offered as if they were binding and current.
The safest assumption when you see a cluster of these signs is not “this is definitely AI,” but “this filing did not get the level of human review our rules require.” That framing matters, because courts sanction conduct, not tools.
Step Two: Fix the Record, Then Decide How Hard to Swing
Your first duty is to your own client and to the court. That means correcting the record—clearly and efficiently—before you worry about scoring style points.
In your response brief, you can:
- Cleanly distinguish or debunk the bad authorities. If a case does not say what opposing counsel claims, show the relevant passage and explain the mismatch. If it is from an inapposite jurisdiction, say so.
- Note the pattern. One bad citation might be a typo. A dozen misrepresentations, non-existent cases, or fictitious quotes tell a different story. Judges are increasingly explicit that patterns of “AI hallucinations” are unacceptable, regardless of intent.
- Tie it to the outcome. Explain how reliance on these flawed authorities would distort the applicable standard, misstate the law, or prejudice your client if accepted.
Sometimes, that is all you need. A judge who has already warned against AI misuse does not require a lengthy technology tutorial; a short, clinical correction will be enough to undermine your opponent’s position.
In other cases—especially where the errors are repeated, serious, or evasively defended—you have a strategic choice: do you ask for sanctions or fees, or do you simply let the judge draw their own conclusions?
When to Seek Sanctions or Fees
Sanctions are a remedy, not a reflex. You should consider them when the opponent’s AI-driven missteps cross certain lines:
- Volume and severity. Is this a single stray case that does not exist, or a brief where “nearly all” the quotes and authorities are fabricated or misrepresented? Courts have been most receptive to sanctions when the misconduct is cumulative and obvious.
- Opportunity to cure. Did opposing counsel promptly acknowledge and correct the problem once it was raised, or did they double down, deflect blame to a “consultant” or junior lawyer, or ignore the issue entirely? Judges have repeatedly said that candor and remediation matter.
- Prejudice and cost. Has your client incurred real expense chasing down fake authorities or dealing with procedurally improper AI-generated filings? Some courts have ordered personal-cost awards payable to the opposing party when lawyers’ AI misuse consumed time and resources.
If you decide to seek sanctions or fees, frame the issue around rules the court already cares about:
- Rule 11 or state analogs: failure to conduct a reasonable inquiry into the law before filing, and presenting legal contentions that are not warranted by existing law or a nonfrivolous argument for its extension.
- Professional-conduct rules: competence, candor toward the tribunal, and misuse of technology without adequate supervision.
- Court-specific AI orders or practice directions: many courts now have standing orders warning that AI-generated citations must be independently verified and that “hallucinated” authorities may result in sanctions.
Your motion does not need to be an AI manifesto. It can be a straightforward application of existing rules to conduct that happens to be mediated by a chatbot. (Note that at least one court ordered a lawyer to pay fees personally when he certified AI-generated briefs without opening the cited opinions)
Using AI Missteps to Challenge Adequacy and Credibility
AI mistakes are not just about citations. In class actions, mass torts, or representative proceedings, they can become evidence that counsel is not an adequate representative of the class. At least one court has found that a lawyer’s failure to correct AI-hallucinated case law—even after being put on notice—contributed to a finding of inadequacy and delayed approval of a class settlement.
More broadly, a lawyer who serves up fake or misquoted cases and then shrugs it off as “the AI’s fault” has a credibility problem. You can use that, carefully, to:
- Undermine reliability. In oral argument or in later phases of the case, politely remind the court that opposing counsel has already put fabricated or misrepresented authorities before the court in this matter. That context can color how the judge receives their factual and legal assertions going forward.
- Resist future shortcuts. If the other side asks for relaxed deadlines, leniency on page limits, or other accommodations, you can argue that their past misuse of AI shows why the court should insist on rigor, not shortcuts.
- Frame settlement dynamics. In some matters, demonstrating that the other side is careless with its legal analysis can subtly strengthen your negotiating posture. They know, and you know, that their credibility with the court has taken a hit.
The key is tone. Judges will not reward gloating, but they do respond to calm, documented demonstrations that an opponent’s conduct has made the litigation more expensive and less reliable than it should be.
Dealing With Pro Se Opponents Who Use AI Badly
Pro se litigants present a different challenge. Courts are generally more forgiving when self-represented parties misuse AI, particularly where there is no clear pattern of abuse or bad faith. Some appellate courts have expressly warned pro se parties that future AI-generated hallucinations may be sanctionable, while declining to impose sanctions in the specific case as a first offense.
When the pro se opponent on the other side has obviously relied on AI:
- Separate compassion from strategy. You can acknowledge that they are trying to navigate a complex system without counsel while still insisting that fabricated authorities and baseless filings not be indulged.
- Focus on gatekeeping, not punishment. Rather than leaping to sanctions, you might ask the court to strike AI-tainted filings, require leave before future submissions, or issue a warning that any further use of fictitious cases will result in specific consequences.
- Use judicial warnings to your advantage. If the court has already issued an order or standing notice about AI, cite it and explain how the pro se filings run afoul of that guidance. You are aligning yourself with the court’s own concern, not asking it to invent a new rule just for this litigant.
There will be cases where sanctions against a pro se opponent are appropriate—repeat abusive filings, clear disregard of warnings, or bad-faith use of AI to harass. But in many situations, your best move is to keep the case on track, not to turn it into a morality play about technology.
Don’t Become the Story
There is one final, ironic risk when you highlight the other side’s AI mistakes: you put AI use in your case squarely on the court’s radar. That is fine—as long as your own house is in order.
Before you file a brief skewering opposing counsel for relying on hallucinated cases, make sure that:
- Your citations are bullet-proof. Run your own “sanction-proof” checklist: every case pulled from a reputable database, read in full, accurately quoted, and correctly characterized.
- Your team knows what tools were used. If you used AI at any point in your drafting process, be prepared to explain exactly how and what verification you did.
- Your tone reflects professionalism. Judges are more likely to act on your concerns if you come across as the adult in the room, not as someone trying to score viral points about “robots practicing law.”
The bigger story in this series is that AI has not changed our core obligations; it has merely created new ways to fall short of them. When the other side does, there is room to use that failure to your client’s advantage—in motions, in fees, in credibility, and sometimes in outcomes. But the power of that argument depends on your ability to show that you are still doing the old-fashioned work: thinking, checking, and standing behind what you file.
Next time, we’ll look at the disclaimers buried in AI terms of service — and why they won’t save users, lawyers, or clients from the consequences of trusting them.
A Word About SilverCain
Silver Cain PLC was founded on the premise that businesses deserve both exceptional litigation experience and direct partner access — and that you should not have to choose between them. Leon Silver and Rebecca Cain have spent decades handling the most complex business and real estate disputes in Arizona and nationally. If you are evaluating counsel in Phoenix, we welcome the conversation.
