When courts sanctioned the attorneys in Mata v. Avianca, Inc. for submitting artificial intelligence (AI)-generated citations to nonexistent cases, they reached for the vocabulary of negligence. That characterization is wrong. Large language models do not forget cases the way attorneys do; they generate citations that look real but correspond to nothing in any reporter, and they do so, as studies have since documented, at rates between fifty-eight percent to eighty-eight percent on legal queries. The tools’ general propensity to fabricate was publicly known by 2023, before most of the incidents now filling the reporters. An attorney who submits AI output without verification does not merely fail a diligence standard. They submit a representation to a court while consciously declining to establish any basis for its accuracy, satisfying the elements of reckless misrepresentation under Restatement (Second) of Torts § 526(b)–(c). The cases that have followed Mata reach instinctively for language that exceeds negligence, calling citations “bogus” and monetary sanctions “insufficient deterrents,” and some have begun to reach for the word “reckless” without grounding that intuition in a cross-regime doctrinal framework. This Essay supplies that framework.

This Essay makes three original contributions. First, it introduces the concept of “structurally predictable error” to explain why AI hallucination is categorically different from human memory failure: the error follows from the tool’s architecture rather than from individual cognitive lapse, its rate is publicly documented, and a reasonable practitioner in the field either knows or should know its hallucination rate before using the tool. Second, it builds the doctrinal case for reckless misrepresentation by connecting Section 526(b)–(c) to the philosophy of AI output, arguing that an attorney who relies on a tool architecturally indifferent to the distinction between truth and falsity adopts that indifference as their own representational posture toward the court. Third, it proposes a graduated three-tier accountability framework that calibrates sanctions to actual culpability, preserving space for good-faith verification efforts and the access-to-justice benefits that AI tools offer to under-resourced practitioners, while refusing to treat willful blindness toward a known category of fabrication as mere carelessness.

Continue reading Hallucinated Authority: AI Citations as Reckless Misrepresentation.