The AI Arms Race in the Courts

Para ler o artigo em português, clique aqui

At the airport, before even collecting their luggage, a passenger receives a message on their phone. The flight was delayed by three hours. A platform offers hundreds of Brazilian reais in exchange for selling the right to compensation. Within minutes, the ticket data are converted into a claim. The case is filed. The proceedings begin. 

Scenes like this have become commonplace. Brazil now registers an unusually high volume of passenger lawsuits against airlines, far above international averages. The phenomenon reflects a system operating at a scale of litigation that is disproportionate by any standard, as documented in the Justice in Numbers 2025 report by the National Council of Justice (CNJ), which consolidates official data on Brazil’s procedural volume. 

When comparing Brazil with EU member states using data from the European Commission for the Efficiency of Justice (CEPEJ), the report reveals a structural imbalance. The rate at which demand is met in Brazil is higher than in Europe. Brazilian judges are, comparatively, more productive—but they face an incoming caseload that has no parallel on the European continent. 

Source: Justiça em Números 2025, Conselho Nacional de Justiça (CNJ)

But the phenomenon has evolved. Consumer litigation has entered a new phase—less visible and more sophisticated. Automated claims are now met with automated responses. It is no longer merely a dispute between lawyers. It is a battle of algorithm against algorithm. Research conducted for the launch of the Chambers Brazil Contentious Guide 2026 confirms this shift. Law firms and inhouse legal teams repeatedly highlighted the central role of technology in managing consumer disputes. 

Institutional recognition of predatory litigation as a systemic problem is relatively recent. In 2024, the CNJ issued Recommendation No. 159/2024, guiding courts to adopt measures to identify and address abusive patterns of mass litigation. The measure formalised what practitioners had already observed: the coordinated repetition of claims with weak factual grounding and high economic pressure. 

For defendants, advantage no longer lies in responding to each claim individually, but in the ability to decipher the logic behind the repetitive demands. Lawyers now confront not only the merits of a single case, but the architecture that produced it. Jurimetrics tools make it possible to map concentration by lawyer, repetition of arguments, anomalous distribution across courts, and documentary inconsistencies. 

Cases like the one that follows are routine. A single firm files dozens of actions per day against the same company. The pattern repeats: real claimants but living in other cities; near-identical narratives. By analysing the full caseload with the support of an AI tool designed for pattern detection, the company finds a high volume of claims filed by the same lawyer, with virtually identical texts. The combined value of the judgments reaches millions. Because the automated response operates at a structural rather than an individual, case‑by‑case level, it is possible to reduce the amount effectively paid by a significant margin. In situations already observed, the final sum disbursed has fallen to approximately 1% of the initial exposure. 

This type of approach mirrors the logic of the Judiciary’s Intelligence Centres, created to identify repetitive claims and anomalous patterns. Courts such as the São Paulo Court of Justice (TJSP) and the Federal District Court (TJDFT) have published technical notes on predatory litigation and abusive claim-splitting, formally recognising the phenomenon. 

Brazil offers particularly fertile ground for this dynamic. Broad, low-cost access to small claims courts, the digitalisation of the judiciary, and a strong consumer-protection culture have lowered the barriers to litigation. Electronic proceedings have removed geographical constraints. As a result, forum selection has become strategic—a practice known as forum shopping. 

Consumer Law Partner, Brazilian Law Firm 

The aviation sector serves as a laboratory for this experiment. Although Brazil accounts for roughly 3% of global air traffic, it is responsible for 95% of the sector’s lawsuits worldwide. A study by Abear monitoring 20,000 flights by Brazilian airlines in Latin America—with a sample of 46% Brazilians and 54% foreigners—found that one in every 362 Brazilian passengers sought some form of judicial redress, compared with one in every 7,432 foreign passengers. 

Litigation rate per passenger in Latin America:

The volume of claims is high, the issues are highly standardised, and individual values are relatively low. It is the ideal environment for testing automated claim-generation models—and now, automated detection of the patterns that sustain them. 

There is an obvious risk. When litigation begins to be treated as a standardised routine, the ability to distinguish mass claims from strategic ones is lost—in a field that also hosts complex and structural debates. 

Consumer Law Partner, Brazilian Law Firm 

Yet this same context creates an opportunity. Data-driven approaches reintroduce nuance where repetition once dominated. In recent cases, courts have recognised predatory litigation patterns, dismissed claims, and imposed bad-faith penalties. The impact of these decisions goes beyond the individual case and sends a clear institutional signal.

It is within this scenario that the metaphor of an AI battle takes on strategic relevance. In the first phase of mass litigation, the advantage lay with those who filed the most claims. In the second, it lies with those who best interpret the patterns to re‑establish distinctions between the legitimate and the fabricated. Interpretation becomes the differentiating factor, reopening space for qualified human intervention and shifting professionals from repetitive work to interpretive work.