Speed is not an excuse. Verification is not optional.
En las salas de justicia de Colombia, la inteligencia artificial ha comenzado a rendir cuentas ante la ley que pretendía servir. La Corte Suprema sancionó a un abogado por presentar citas judiciales inventadas generadas por una herramienta de IA, marcando la primera vez que el poder judicial colombiano impone consecuencias formales por el mal uso de estas tecnologías en litigios. El caso no es un accidente aislado, sino el reflejo de una tensión más profunda: la velocidad que prometen las máquinas frente a la responsabilidad que exige la justicia. En un mundo donde los algoritmos pueden fabricar autoridad con la misma facilidad con que la citan, los tribunales del mundo comienzan a recordarle a la profesión jurídica que la verdad no es opcional.
- Un abogado presentó ante la Corte Suprema una petición de revisión extraordinaria con citas de fallos que nunca existieron, generadas por una IA que inventó jurisprudencia con aparente confianza.
- Cuando el tribunal señaló los errores y exigió correcciones, el abogado reincidió: el nuevo escrito contenía las mismas fuentes falsas más referencias adicionales a artículos inexistentes, agravando la falta en lugar de remediarla.
- Solo ante una segunda confrontación el abogado admitió haber usado IA para acelerar su trabajo, revelando una práctica extendida en la profesión que hasta ahora operaba sin consecuencias visibles.
- La Corte Suprema impuso una multa equivalente a 15 veces el salario mínimo mensual, dejando en claro que la responsabilidad de verificar lo que se presenta ante un tribunal no puede delegarse en ningún algoritmo.
- El caso se suma a un precedente previo en el que el mismo tribunal anuló la sentencia de un juez que usó IA para redactarla, después de descubrir que el fallo contenía resúmenes de pruebas falsas y testimonios de audiencias que nunca ocurrieron.
- Colombia se une así a una tendencia global de sanciones judiciales por fuentes legales fabricadas por IA, señalando que el umbral de tolerancia institucional ha llegado a su límite.
La Corte Suprema de Colombia trazó una línea que el sistema judicial había evitado dibujar durante años. En noviembre de 2025, sancionó a un abogado con una multa de 15 veces el salario mínimo mensual por presentar documentos legales con citas inventadas y referencias falsas generadas por inteligencia artificial.
El abogado había recurrido a una herramienta de IA para redactar una petición de revisión extraordinaria —un recurso de alto impacto— con el propósito de agilizar su trabajo. La herramienta alucinó: fabricó fallos inexistentes, distorsionó artículos legales y construyó una arquitectura jurídica que no tenía sustento real. Cuando la Corte señaló los errores y solicitó una corrección, el nuevo escrito reprodujo las mismas fuentes falsas y añadió referencias adicionales a artículos que no correspondían a lo citado. Solo ante una segunda confrontación el abogado reconoció el origen del problema.
La respuesta del tribunal fue contundente. Los magistrados subrayaron que la obligación de verificar la exactitud de lo que se presenta ante un tribunal de última instancia no admite excepciones, y que la velocidad no puede ser excusa para comprometer la integridad del proceso.
Este caso no surgió en el vacío. Meses antes, la misma Corte había anulado la sentencia de un juez que utilizó IA para redactarla, tras descubrir que el fallo contenía resúmenes falsos de pruebas y referencias a testimonios de audiencias que nunca tuvieron lugar. La libertad de un acusado había quedado suspendida sobre una ficción algorítmica; el tribunal revocó la decisión y restableció el derecho a un proceso justo.
Colombia se suma así a una corriente internacional que ya ha comenzado a sancionar a abogados por presentar precedentes inventados y estatutos fantasma. El mensaje que emerge de estos casos es el mismo en todos los idiomas: las herramientas de IA pueden redactar con autoridad aparente, pero la responsabilidad de distinguir lo verdadero de lo fabricado sigue siendo, irrenunciablemente, humana.
Colombia's Supreme Court has begun drawing a line. For years, artificial intelligence slipped into courtrooms with little oversight—judges using it to draft sentences, lawyers deploying it to build arguments. The judiciary tolerated it, mostly in silence. But in November 2025, the court issued a ruling that marked a turning point: it fined a lawyer 15 times the monthly minimum wage for submitting legal documents riddled with fabricated citations and false case references.
The case began simply enough. A lawyer filed an extraordinary review petition—a high-stakes legal motion—using AI to help draft it. The system, as these systems do, hallucinated. It invented case citations that did not exist. It misquoted legal articles, pulling references from thin air. When the Supreme Court flagged the errors and asked for corrections, the lawyer submitted a revised version. The new filing contained the same invented sources, now compounded by additional false references to articles that lacked the numerals being cited and addressed entirely different legal matters.
Only when confronted a second time did the lawyer admit the truth: an AI tool had generated the problematic passages. He had been using it, he explained, to speed up his work—to move cases along faster. The court's response was unsparing. In its ruling, the justices noted that the lawyer had violated the fundamental duty of accuracy and verification that the profession demands. This was not a technical glitch to be overlooked. This was a lawyer presenting fabricated law to a court of last resort.
The Supreme Court's action did not emerge in isolation. Months earlier, the same court had revoked a trial judge's sentencing decision after discovering the judge had drafted it using AI. That sentence contained false summaries of evidence and mischaracterized testimony from hearings that had never actually occurred. A defendant's liberty hung in the balance, and the AI had corrupted the record. The court threw out the ruling and restored the defendant's right to a fair proceeding.
What makes Colombia's move significant is its timing and its clarity. Other countries have already begun sanctioning lawyers for submitting AI-fabricated legal sources—false precedents, invented statutes, phantom case law. Courts in multiple jurisdictions have issued guidelines to prevent spurious sources from entering the record. But Colombia's Supreme Court has now joined that enforcement effort, making explicit what was previously implicit: lawyers are responsible for what their tools produce, and courts will hold them accountable.
The lawyer in this case faced a choice point that many in his profession now face. AI can accelerate work. It can draft quickly. But it can also invent with confidence, generating text that sounds authoritative while being entirely false. The court's message was clear: speed is not an excuse. Verification is not optional. The duty to present truthful law to the court is not something a tool can relieve you of.
What unfolds next will likely follow the pattern already visible elsewhere. As more lawyers experiment with AI, more errors will surface. Some will be honest mistakes. Others will be negligence. Courts will respond with sanctions, with rules, with consequences. The profession is learning, in real time, that artificial intelligence in the courtroom is not a neutral tool—it is a tool that requires vigilance, skepticism, and the kind of careful human judgment that no algorithm can replace.
Citações Notáveis
The lawyer admitted the errors were generated by an AI program he used to expedite case work and procedures— The lawyer's explanation to the court
Multiple jurisdictions have already imposed sanctions on lawyers who submitted documents with non-existent legislation or fabricated precedents, and have issued directives to prevent false sources from entering court proceedings— Colombia's Supreme Court, citing international precedent
A Conversa do Hearth Outra perspectiva sobre a história
Why did the court treat this case as such a turning point? Lawyers make mistakes all the time.
Because this wasn't a mistake—it was a system producing false law with confidence, and a lawyer not catching it. The court saw a pattern: AI inventing citations, the lawyer not verifying them, then doing it again in the correction. That's negligence at scale.
But the lawyer said he was just trying to work faster. Isn't that understandable?
The court didn't buy it. There's a difference between using a tool to draft faster and outsourcing your duty to verify what you submit. The moment you hand something to a court, you're vouching for its truth. You can't delegate that responsibility to an algorithm.
What about the earlier case with the judge? That seems even worse.
It is. A judge drafting a sentence with AI that fabricated evidence summaries—that's not just negligence, that's a violation of someone's right to a fair trial. The court revoked it entirely. That's why they're being so strict now. They've seen what happens when AI gets into the courtroom unchecked.
So what's the real risk here?
That courts become unreliable. That law itself becomes unstable if the sources cited don't actually exist. Once you allow fabricated precedent into the record, you've corrupted the foundation of the entire system.