El Salvador's Constitutional Chamber Sets Ethical and Legal Standards for AI Use in Courts and Public Institutions

By Edy Portal, Eproint

The Constitutional Chamber of El Salvador's Supreme Court of Justice has issued a set of ethical and legal criteria governing the use of generative artificial intelligence within state institutions and the country's courts. The Judicial Branch announced the decision in a statement, explaining that the criteria emerged from a ruling on an unconstitutionality proceeding, case number 57-2025, in which the court identified that the plaintiffs had used an AI tool to help draft their legal filing.

According to the published resolution, the court identified a passage in the complaint that it considered evidence a generative AI tool had been used to build the legal argument. The excerpt, reproduced in the ruling, referenced integrating comparative analysis with the original content and incorporating constitutional and legal provisions from other countries. The chamber treated the passage as sufficient basis to conclude that the challenge had likely been prepared with the assistance of such a tool.

In its decision, the court stated that the central premise for using AI in the administration of justice is respect for fundamental rights. It cited guidance from United Nations bodies holding that the right to an independent and impartial tribunal requires access to a human judge, in the same way that the right to counsel of one's choosing requires access to a human lawyer. The ruling also referenced international jurisprudence, including a precedent from Colombia's Constitutional Court, which had previously established standards for judicial offices using AI tools, among them transparency, accountability, privacy, non-substitution of human reasoning, verification and human oversight.

The Salvadoran court adopted a similar framework, setting out principles that include transparency, accountability, privacy, human control, risk prevention, equality, verification, ethics and adherence to good practices for the use of AI across public institutions and the judiciary. Regarding the specific case before it, the chamber found that the conduct of the plaintiffs conflicted with the principles of transparency, explainability and human oversight that the court expects in filings before it. The underlying constitutional challenge, which sought to have a provision of the Special Law for the Intervention of Telecommunications declared unconstitutional, was ultimately dismissed on procedural grounds; the challenged provision allows telecommunications interception and recording when authorized by one of the legitimate participants in the communication.

El Salvador's ruling adds to a broader pattern of Latin American judiciaries addressing the use of AI tools in legal proceedings. Colombia's Constitutional Court set out its own criteria for judges using AI systems such as ChatGPT in prior rulings, including an order to the country's Superior Council of the Judiciary to issue formal guidelines on the technology's use in judicial offices. In Paraguay, the Supreme Court of Justice approved an internal policy in 2025 establishing directives for the use of AI systems applicable to judges and judicial personnel. El Salvador's Supreme Court had also held a conference earlier this year on the use of AI within its contentious-administrative jurisdiction, reflecting continued institutional engagement with the technology beyond the specific case that prompted the new criteria.

The Salvadoran resolution does not establish a binding statute or regulation but sets interpretive criteria that the Constitutional Chamber says should guide the use of generative AI across public institutions and tribunals going forward. How these criteria will be applied in future cases, and whether the Legislative Assembly or other bodies will develop more detailed regulation on AI use in legal practice, remains to be seen.