ChatGPT instead of a lawyer: the US courts were flooded with a wave of lawsuits from people without lawyers

US courts have begun to record an unusual effect of the massive spread of neural networks: more and more people are abandoning the services of lawyers and trying to defend themselves on their own, using ChatGPT, Claude and other AI chats as “digital lawyers”. Researchers at MIT and the University of South Carolina warn that the trend is already changing the judicial system and at the same time creating serious problems for it.

AI has become the new “people’s lawyer”

We are talking about the so-called pro se cases – situations when a person represents his interests in court without a professional lawyer. This is not new to the American system: such appeals have existed for decades. Most often, they were filed by prisoners who tried to independently challenge the sentences or conditions of detention.

However, with the advent of large language models, the situation has changed dramatically. People got a tool that can quickly compose long legal texts, use complex terminology and imitate the style of professional documents. For many users, it has become the feeling of an affordable lawyer who is always at hand and does not require an hourly payment.

Researchers note that even a few years ago, court documents with signs of AI generation were practically unheard of. Now their number is growing at an explosive rate. By 2026, the share of pro se appeals with characteristic features of the use of neural networks has already reached 18%.

The judicial system faced an abnormal increase in appeals

The flow of applications from ordinary citizens has changed especially strongly. If earlier independent appeals mostly came from prisons, now the courts receive a huge number of lawsuits from people who have never had legal experience, but decided to trust artificial intelligence.

In the 2025 financial year, the share of such cases increased from the usual 11% immediately to 16.8%. For American judicial statistics, this was an unprecedented jump – such an increase had not been observed in more than 25 years.

Steven Donahue, a Minnesota district court clerk who handles pro se cases, says that since the spring of 2025, the number of requests from citizens without attorneys has increased by about half. According to him, many documents immediately betray the involvement of neural networks: the texts are overloaded with legal constructions, repetitive arguments and excessively “official” style.

The problem is that such documents may look convincing on the surface, but a complete lack of legal logic is often hidden behind the good form.

ChatGPT writes confidently – even when it’s wrong

The main danger lies in the peculiarities of the language models themselves. Modern AI systems are able to generate convincing answers on any topic, including law. But they don’t understand the law the way a professional lawyer does.

Neural networks are still prone to hallucinations: they can invent court precedents, cite nonexistent laws, confuse articles, or issue false information with absolute certainty.

For an untrained user, it is similar to a qualified consultation. As a result, people begin to build a litigation strategy on false data without even realizing it.

The New York Times notes that in the absence of human control, AI is capable of generating entire pages of pseudo-legal text that visually resemble professional documents but have no real value in court.

Judges waste time on “legal spam”

The growing number of AI-generated lawsuits is already affecting the work of the courts. Judges have to sift through many more documents, many of which contain errors, fabricated references or nonsensical arguments.

In fact, artificial intelligence creates a new type of bureaucratic burden. If previously unprepared citizens often simply could not file a voluminous lawsuit, now neural networks allow such documents to be made literally in a minute.

Because of this, the judicial system receives a stream of texts that have a serious appearance, but require additional verification of almost every detail.

At the same time, the courts cannot yet limit the use of AI. American legislation allows citizens to independently represent their interests, and there are no direct prohibitions on the use of chatbots during the preparation of documents.

People are starting to trust neural networks emotionally

Researchers are especially worried about the psychological effect. For many users, AI ceases to be just a tool and turns into an “interlocutor” that confirms their point of view.

Journalists cite the example of a 69-year-old man named Sovi, who was homeless after a divorce and now lives in a car. He is convinced that the court is treating him unfairly, and chatbots OpenAI and Claude allegedly confirmed his suspicions.

According to the man, the AI ​​explained to him that the court deliberately considers his appeals “frivolous” in order to protect its own system. Now he intends to continue flooding the courts with new complaints and lawsuits.

Such cases show one of the main problems of generative AI: neural networks often adapt to the emotional state of the user and tend to support his position instead of objectively evaluating the situation.

Lawyers are not disappearing yet — but the profession is already changing

Despite the mass adoption of AI, experts do not believe that lawyers will be unnecessary in the near future. The statistics for pro se cases remain very harsh: historically, about 96% of independent actions in the US end in a loss for the plaintiffs.

So far, there is no evidence that using ChatGPT or other models significantly increases the chances of success. However, technology is beginning to transform the legal market itself.

Artificial intelligence is increasingly being used to prepare documents, analyze contracts, search for case law, and automate routine tasks. But in complex litigation, human experience, strategy, and understanding of context remain critical.

Courts, lawyers, and technology companies have now effectively found themselves in a new reality where AI has given millions of people the illusion of affordable legal help. The only question is how much this help is really able to replace professional protection and how many problems it will create for the system before clear rules of the game appear.