This is some text inside of a div block.

Abusive litigation: how Enter's AI agents detect fraud and violation patterns

Time Enter

Between 2020 and 2024, Brazil's National Council of Justice (CNJ) recorded more than 129,000 lawsuits citing abusive litigation. In the first half of 2025 alone, 277 new lawsuits were filed every day.

This is no coincidence. In 2022, the lawsuit filing rate per 100,000 people started growing exponentially — an unprecedented shift in Brazil's judicial system that coincided with the mass adoption of ChatGPT in the country.

AI entered the legal industry on two fronts simultaneously: the same technology driving mass lawsuit filing now lets companies detect, at scale, fraud patterns and violations that traditional methods would rarely uncover.

Gráfico animado: processos novos por 100 habitantes no Brasil — litigância abusiva

For companies, the impact of rising litigation is both financial and structural. Brazil's seven largest banks, for example, hold a combined US$15 billion in provisions for legal proceedings. The same pattern repeats across industries.

National Diagnosis on Combating Abusive and Predatory Litigation in the Judiciary, CNJ 2025

Hidden patterns made visible by Enter's AI

Operating across the litigation portfolios of Brazil's largest companies, Enter's technology has mapped the country's most recurring patterns of fraud and violation.

Enter's AI agents run more than 30 checks per lawsuit, pulling from court records, internal documents and prior litigation history, such as:

  • Coordinated litigant networks
  • Document tampering
  • Standardized filings
  • Forum shopping
  • Collusion among legal professionals

One pattern goes beyond conventional abusive litigation, and has only recently received media coverage: prompt injection — a technique where instructions are hidden inside an initial filing — written in a reduced font or a color that blends into the background — to manipulate whatever AI system a judge or company uses to read the lawsuit, nudging it to favor whoever planted the prompt.

At Enter, we've found this type of incident across several industries. At a financial institution, for example, Enter's technology identified more than 80 lawsuits with prompt injection in filings from a single attorney. The instruction was identical across all lawsuits and appeared — at a smaller scale — in other banks' portfolios, ruling out the possibility of an isolated incident.

In the healthcare industry, Enter detected more than four explicit commands directed at the AI used by the judge in a single lawsuit, totalling a claim value exceeding US$18,700.

Scan horizontal de petições por agentes de IA da Enter

Monitoring these lawsuits yielded a concrete result for the legal department of one of Enter’s clients: the company obtained a ruling that expressly recognized the practice. In the decision, the judge identified the prompt injection, which contained a direct instruction to AI systems to grant requests for a fee waiver, emergency relief and a defendant summons. The judge dismissed the lawsuit without prejudice and ordered notifications to the Brazilian Bar Association and the Public Prosecutor's Office to investigate possible criminal wrongdoing.

Across every industry and jurisdiction in Brazil, the same pattern repeats: fraudulent lawsuits only become visible when thousands of them are analyzed together.

Below, we show how this phenomenon plays out in practice — with real examples of fraud and violations Enter’s AI identified at financial institutions, aviation companies, technology firms, healthcare operators, and telecommunications providers.

How abusive litigation manifests by industry

Financial Institutions

For financial institutions, abusive litigation is not confined to a single product. It spans from payroll-deductible credit to checking accounts. The challenge for companies’ legal departments is delivering the right legal inputs for every lawsuit, at scale.

In payroll-deductible credit, volume is the first obstacle: the product concentrates the highest litigation rate in Brazil, with more than 1,700 new lawsuits filed everyday. Financial institutions' daily ordeal is to identify plaintiffs who file malicious suits, pinpoint the means used in such claims, and adjust claim-response operations proactively.

casE #1

Working with Enter, the legal department of Banco Mercantil identified a single plaintiff who filed hundreds of lawsuits against the institution in less than 10 months. Through the noise, Enter's AI agents identified a pattern: 97% of lawsuits included identical hearing waiver requests, and 99% included fee waiver requests without justification to support it.

We were able to trace a client acquisition company that was funneling plaintiffs into an abusive litigation network targeting Mercantil. After partnering with Enter, the legal team of the bank secured more than 40 dismissals without prejudice, two malicious prosecution fines applied to the opposing party, and a 98% drop in lawsuits filed by the same litigant in the following quarter.

Fraud and wrongdoing detection are one of the factors that drive measurable outcomes for large companies. With Enter, Mercantil achieved a 12 percentage point increase in legal department portfolio performance and a 25% reduction in operating costs.

casE #2

In Nubank's portfolio, Enter's AI mapped a network of 10 attorneys responsible for 38% of lawsuits — using fraudulent documents and identical filings.

Exposing these malicious prosecutions was part of a broader partnership that strengthened Nubank's legal team, resulting in a 6 percentage point increase in legal department portfolio performance across the three main consumer litigation subjects, measured across lawsuits filed in the same month.

casE #3

At another bank, the client's internal team had already flagged more than 220 attorneys with suspicious behavior. After the AI checks, the number doubled. When Enter mapped attorney networks operating in coordination, it found more than 470 abusive plaintiffs. The same analysis found:

  • 165 attorney networks
  • 3,411 identical powers of attorney
  • 206 shared addresses
  • 834 cross-referrals among the parties involved
Animação ilustrando rede de advogados e escritórios parceiros da Enter

casE #4

Em uma outra instituição, irregularidades também foram encontradas em documentos. Em um dos casos, o comprovante de residência apresentado como prova era uma adulteração do documento original do próprio advogado que assinou a petição. Mesmo código fiscal, mesmo valor, mesmos dados de conta. O tipo de correspondência que só aparece quando a IA lê centenas de arquivos ao mesmo tempo.

Aviation

Brazilian airlines faced more than 365,000 lawsuits in 2025 — a 32% year-over-year increase, compared to 24% in financial services and just 1% in telecommunications.

LATAM's operations in Brazil represent roughly 50% of the group's activity but account for more than 98% of the company's lawsuits. In 2025 alone, the airline faced more than 120,000 passenger-related lawsuits.

At this volume, manually reviewing each flight's context and building the right legal inputs for every lawsuit answer wasn't feasible. That was one of the drivers behind the partnership between LATAM and Enter.

LATAM built one of the key workstreams on Enter's technology: fraud and violation detection. To that end, Enter needed to ensure the most robust information concentration was made available to LATAM’s legal team to support each defense type. Each lawsuit became enriched by AI with external and operational data tied to that specific flight, including:

  • Weather conditions
  • Operating conditions at every airport in the country
  • Evidence of disruptions affecting other airlines operating at the same airport on the same date

For lawsuits regarding trip cancellations, for example, AI agents verify the full context at the time of the flight and cross-reference it with operational records to determine whether the cancellation was justified.

As a result, Enter’s technology identified systemic fraud across the portfolio:

  • Proof of residence reused across dozens of lawsuits filed by different plaintiffs
  • Powers of attorney with no valid signature
  • Deceased plaintiffs
  • Attorneys with suspended bar registration

In one instance, reusing the same proof of residence resulted in dismissal of the lawsuit, a malicious prosecution fine of equal to 9% of the claim value imposed on the plaintiff, and the opening of a criminal investigation against the sponsoring attorney.

In less than twelve months,  the legal department of the airline recorded a 13% reduction in operating costs and saved US$2.8 million with the support of Enter’s platform. Driven by the financial impact, LATAM expanded the portfolio supported by Enter from 12% to 70% of its consumer litigation.

Technology and Retail

Enter identified specific malicious prosecution litigation patterns in e-commerce platforms and technology companies, including forum shopping and abusive claim splitting in digital identity theft schemes, enabled by the digital nature of the sector.

case #1

In one partner supported by Enter’s technology, 15% of lawsuits had a detectable violation. In that portfolio, agents identified plaintiffs filing in states with more consumer-friendly case law and tampered with evidence to support their claims of territorial jurisdiction, such as:

  • Proof of residence in someone else's name
  • Proof of residence issued more than 6 months prior
  • Expired power of attorney

case #2

In another portfolio, our AI identified a pattern of abusive claim splitting in lawsuits related to digital identity theft. When fake accounts impersonate an attorney to extort clients, the affected party may bring a claim against the platform.

Instead of consolidating all incidents into a single lawsuit, some plaintiffs file separate lawsuits for each falsified profile — to multiply fines from injunctions and stack moral and material damages claims.

By cross-referencing the origin of the lawsuits, plaintiffs and filing history, Enter’s AI flagged suspicious claim splitting in 10% of the portfolio, driving a 10 percentage point increase in legal-department portfolio performance over the prior period, and supporting the legal team and the firms representing the client company with legal inputs that back the fraud strategies presented to the judge.

Healthcare

case #1

In healthcare portfolios, fraud reaches a different level of organization, involving coordinated operations between physicians and plaintiffs' attorneys.

An Enter study analyzed the lawsuit database of a health insurer in search of potential collusion among these professionals. Enter identified a single attorney behind more than 3,800 lawsuits — with evidence of collusion across a network of 140 physicians. The analysis answered the following questions:

  • Attorney and physician: are they working together?
  • Volume: in how many lawsuits?
  • Medical reports: are they generic and focused on the same subject?
  • Specialty: does the physician hold a valid medical registration and a specialty compatible with the lawsuit?
  • Proof of address: are there signs of document tampering to influence requests for out-of-network treatment or treatment at specific clinics?

Among physicians who most frequently provide psychiatric reports for one recurring attorney, the AI revealed that three of the top five do not hold the specialty required to issue that medical report — a practice considered an ethical violation by Brazil's Federal Council of Medicine.

case #2

Similar fraud patterns emerged in the partnership with SulAmérica. Across a portfolio of 30,000 lawsuits, Enter’s AI identified a single attorney responsible for more than 1,000 lawsuits — many with identical medical reports frequently issued by the same medical professionals. Read the full details in the customer story.

Telecommunications

The telecom sector concentrates high volumes of lawsuits in small claims courts, with standardized petitions and small groups of attorneys holding disproportionately large lawsuit portfolios.

At one operator, Enter’s technology mapped more than a thousand lawsuits filed by a small group of attorneys — all containing requests related to Brazil's data protection law and demands for historical records of phone lines, even where plaintiffs no longer had an active contract with the operator. Cross-referencing with Enter Network, agents identified:

  • Identical structure across the lawsuits
  • Filing dates concentrated within narrow time windows
  • Shared addresses among formally distinct plaintiffs

In one lawsuit where pro se litigation was not allowed, the same attorney appears as both plaintiff and sponsoring attorney in separate filings. In another analysis, Enter found that the same proof of residence linked to one of the attorneys appeared in more than 20 lawsuits filed by different individuals.

The work also found consumer reviews on Reclame Aqui and Consumidor.gov.br (Brazil's two main platforms for resolving disputes with companies) being used as proof of prior resolution attempts — with nearly identical language across filings, pointing to bulk templating.

Across every industry, fraud and violations only become visible when thousands of lawsuits are analyzed together. These are some of the patterns we detected. With Enter's AI, this identification now happens at scale. Instead of tackling each lawsuit in isolation, the company’s legal department proactively identifies patterns, brings the strongest arguments to judges, and adds concrete evidence to the record — building a stronger defense against fraud systemically.