When AI invents fake legal citations, does it cost solo lawyers more than big law firms, or are they just easier to catch?

This explores whether solo and small-firm lawyers are punished differently from big-firm lawyers when AI errors show up in court filings. The short answer is that the collection can't settle the question directly. It has two pieces that point in an unexpected direction, though: the real difference by firm size may be in who gets caught, not in who gets punished.

The first piece is about who shows up in the record. In a set of 114 US court cases with suspected AI errors, 90 percent involved solo or small firms, and more than half involved plaintiff's counsel Do small law firms misuse AI more often than large ones?. That looks like a story about small firms misusing AI. The note says plainly that it isn't one: these are errors that were detected, not a measure of how often each kind of firm actually misuses AI. Large firms have more layers of review before anything is filed, and their mistakes may be caught internally and never reach a judge. So the 90 percent figure tells you whose errors become visible, not whose errors exist.

The second piece is about what happens once an error is found, and the answer is often very little. Courts that found fabricated or suspected AI citations frequently imposed no penalty aimed specifically at the AI error Do courts actually sanction fabricated AI citations when detected?. When sanctions did come, they depended on the judge's discretion, on whether anyone was actually harmed, and on whether the lawyer seemed to act in bad faith. The fake citation alone usually wasn't enough. That matters for the firm-size question. If punishment depends on harm and intent rather than on the error itself, any gap between solo and big-firm outcomes would come through those judgment calls, such as how believable a lawyer's explanation sounds or how much they've cost the other side. It wouldn't come from a fixed rule. The collection doesn't measure whether those judgment calls favor one kind of firm.

There's a useful parallel from outside law. A randomized experiment on AI rules for academic peer review found that banning AI use versus allowing limited use barely changed review outcomes, while large shares of reviewers broke whichever rule they were given Does banning LLM use in peer review change review outcomes?. Courts look similar. The rules on paper matter less than detection and discretion, so the useful question isn't which firms get sanctioned more harshly but which firms' mistakes become visible at all.

To be clear about the gap: no note here directly compares sanction outcomes for solo lawyers and large firms. The firm-size study counts incidents, and the sanctions database looks at a small number of cases without breaking them down by firm size. A real answer would need the two combined: sanction outcomes sorted by firm size, ideally with some way to estimate the errors that never reached a judge.