The Paid Witness Has Always Been in the Courtroom: How Legal Systems Keep Losing the Same Argument

In 2023, a federal judge in the Southern District of New York issued a sanctions order against two attorneys who had submitted a brief citing legal precedents that did not exist. The citations had been generated by an artificial intelligence tool that, when asked to produce supporting case law, had simply invented it. The attorneys had not verified the citations before filing. The judge's opinion described the incident as unprecedented. Legal commentators called it a new category of professional failure, a product of emerging technology and insufficient caution.

Photo: Southern District of New York, via www.justice.gov

It was, in fact, a very old problem wearing new clothes. The specific mechanism was novel. The underlying dynamic — a formal legal proceeding corrupted by testimony that was not what it claimed to be — has been documented continuously for approximately four thousand years. Every civilization that developed courts developed, almost simultaneously, the problem of manufactured authority. The solutions have been creative. None of them has worked.

Mesopotamia and the Original Evidentiary Crisis

The law codes of ancient Mesopotamia are among the earliest written legal documents in human history, and they reveal something striking: the drafters were already worried about witnesses. The Code of Hammurabi, dating to roughly 1754 BCE, contains explicit provisions against false testimony, including a remarkable clause specifying that a witness who gave false evidence in a capital case would himself be subject to the death penalty.

Photo: Code of Hammurabi, via kajabi-storefronts-production.kajabi-cdn.com

The severity of the prescribed punishment is itself informative. Lawmakers do not generally draft death penalties for behaviors that are rare or theoretical. The false witness clause in Hammurabi exists because false witnesses were common enough to constitute a recognized problem requiring a formal deterrent. The courts of Babylon were, within a generation of their establishment, already contending with testimony shaped by payment, loyalty, or self-interest.

Assyrian court records from the second millennium BCE include cases in which witnesses were explicitly identified as having been brought to court by one of the parties — what we would today call party-sponsored testimony. The records do not describe this as inherently improper. They describe it as a normal feature of proceedings that the judges were expected to weigh accordingly. The ancient Mesopotamian legal system did not eliminate the paid witness. It acknowledged him and built procedures around his predictable behavior.

Rome's Attempt at Regulation

Roman law approached the testimony problem with characteristic systematizing ambition. The Digest of Justinian, compiled in the sixth century CE but drawing on centuries of prior legal development, contains extensive discussion of witness credibility — specifically, of the factors that should lead a judge to discount or disregard testimony.

Photo: Digest of Justinian, via is1-ssl.mzstatic.com

The list of disqualifying characteristics is illuminating: slaves (whose testimony was considered inherently coerced), people with a financial interest in the outcome, individuals of low social standing, those with a prior record of dishonesty, and — with particular relevance — anyone who had a close relationship with the party on whose behalf they testified. Roman jurists were not naive about the connection between social obligation and courtroom narrative. A man who testified favorably for his patron was not, in their framework, simply a loyal subject of the court. He was a man whose testimony had been purchased in the currency of social dependency.

The Roman solution was to weight testimony by the social status and independence of the witness — a system that reduced some forms of purchased testimony while creating a different problem: wealthy, high-status witnesses were presumptively credible regardless of their actual relationship to the facts. The reform addressed one vector of corruption and institutionalized another.

The Expert Witness and the American Innovation

American courts in the nineteenth century confronted a new version of the testimony problem, one that earlier legal systems had not fully anticipated: the rise of technical knowledge that judges and juries could not independently evaluate.

The expert witness — a person compensated to offer specialized opinion rather than factual observation — was a reasonable response to genuine complexity. Industrial accident cases required engineers. Medical malpractice cases required physicians. Patent disputes required people who understood the relevant technology. The courts needed experts, and experts needed to be paid. The arrangement was logical.

What the arrangement also created was a market. By the late nineteenth century, American litigation had developed a recognizable class of professional expert witnesses — individuals who appeared repeatedly on behalf of parties who could afford their fees, offering opinions that were, reliably, favorable to their clients. The term "hired gun" entered legal discourse during this period, and it was not complimentary. The expert witness had become, in the popular understanding, simply a more expensive and credentialed version of the false witness that Hammurabi had tried to deter with capital punishment.

The Federal Rules of Evidence, substantially revised in the 1970s and updated repeatedly since, represent the most sustained American attempt to manage this problem. The Daubert standard, established by the Supreme Court in 1993, requires federal judges to act as gatekeepers — evaluating the methodology behind expert testimony before allowing it to reach a jury. The standard was designed to exclude junk science. Studies conducted in the decades following Daubert found that it reduced the volume of expert testimony somewhat and increased the cost of litigation substantially. Whether it improved the accuracy of verdicts remains actively debated among legal scholars.

The Digital Testimony Problem

The 2023 AI citation case is, in this light, not a departure from the history of paid or manufactured testimony. It is its latest iteration. The attorneys did not intentionally fabricate authority. But they submitted, under their professional signatures, content they had not verified — content generated by a system optimized to produce plausible-sounding output rather than accurate information. The mechanism was different from a Babylonian witness paid to misremember. The structural problem was identical: something entered the formal record that was not what it represented itself to be, and the court had no reliable way to detect it without external investigation.

Legal technology companies are currently developing AI-based tools for detecting AI-generated fabrications in legal filings. This is, in the most precise historical sense, a new mechanism for an ancient arms race. Mesopotamian courts had professional investigators to challenge witness credibility. Roman courts had the advocatus whose job included discrediting opposing witnesses. American courts have Daubert hearings and, now, algorithmic verification tools. Each generation builds the next layer of scaffolding around the same structural vulnerability.

The Persistent Human Constant

The reason this problem has never been solved is not that legal systems have been insufficiently clever. Roman jurists were extremely clever. The drafters of the Federal Rules of Evidence were working with centuries of accumulated legal theory. The reason the problem persists is that it is not, at its root, a legal problem. It is a behavioral one.

People with a stake in an outcome will shape their testimony toward that outcome. This is not a moral failing unique to corrupt individuals. It is a description of how human cognition operates under conditions of self-interest — a phenomenon that has been documented in controlled psychological experiments, in the historical record, and in every courthouse that has ever convened a proceeding. The legal systems that have managed it best are not those that eliminated the incentive to shade testimony. They are those that built procedures robust enough to function despite the certainty that someone in the room is doing exactly that.

The clay tablets of Babylon and the docket of the Southern District of New York are, on this particular question, the same document.