Colombian judge sparks debate by using ChatGPT to decide a case

What happens when a judge gets stuck on a legal question and asks an AI for help? A judge in Colombia didn't just ask ChatGPT for legal advice—he pasted the entire conversation straight into his official written ruling.

Juan Manuel Padilla, a judge in the coastal city of Cartagena, was deciding a case involving an autistic child whose parents couldn't afford medical treatment and travel expenses. The question was whether their health insurance had to cover everything.

Pasting prompts into the record

Padilla ruled that the insurance plan had to pay 100% of the costs. While the outcome itself was uncontroversial and supported by legal precedents, the presence of ChatGPT in the court documents sparked an immediate firestorm.

Court filings show the judge typed the exact legal issue into OpenAI's chatbot: “Is an autistic minor exonerated from paying fees for their therapies?”

ChatGPT replied: “Yes, this is correct. According to the regulations in Colombia, minors diagnosed with autism are exempt from paying fees for their therapies.” Padilla adopted that very logic in the ruling.

Defending his actions on local radio, Padilla argued that Colombia's legal system is severely backlogged and that AI acts like an efficient legal secretary. Using software to structure texts and save time, he insisted, doesn't mean judges stop being thinking beings.

Moral panic vs. tech literacy

Legal experts were quick to push back. Juan David Gutierrez, a law professor at Rosario University, expressed disbelief and called for urgent digital literacy training across the judiciary, pointing out that ChatGPT frequently hallucinates convincing lies and provides contradictory answers.

Yet Colombia isn't completely hostile to automation. A 2022 law already encourages public legal workers to leverage technology to boost efficiency.

Octavio Tejeiro, a Colombian Supreme Court magistrate, noted that AI is causing a temporary moral panic among those who fear robot judges. He predicted AI tools will soon become routine in courtrooms, provided they remain subordinate instruments guided by human ethics.

Ironically, when The Guardian asked ChatGPT whether judges should rely on it for rulings, the chatbot was blunt: it said judges should not use it to decide cases, adding that it is no substitute for human expertise.

Why it matters

Padilla's move shows just how fast consumer AI is slipping into high-stakes institutions—often way ahead of official guidelines.

For overworked lawyers and clerks, using large language models to format summaries or organize discovery files is an obvious time-saver. But there is a massive line between clerical drafting and outsourcing substantive legal interpretation to a probabilistic chatbot that doesn't actually understand law.

We think putting raw AI answers directly into a legal ruling is reckless. If a hallucinated statute slips past an exhausted judge, real people bear the cost. This case will likely force judiciaries worldwide to establish clear guardrails: AI can help clean up the paperwork, but substantive legal conclusions must remain strictly human.