The Bottom Line
- AI is an assistant, not a replacement. The core principle is effective human control. All judicial decisions, from assessing evidence to the final ruling, must be made by a human judge, ensuring accountability remains with the judiciary, not an algorithm.
- Restricted tools and data usage. Judges are limited to using government-approved, vetted AI tools. Crucially, they are banned from using AI for profiling individuals, risk assessment, or handling specially protected personal data within their judicial functions.
- No ‘black box’ decision-making. While AI-generated drafts are permitted as support tools, they require “complete and critical” review by the judge. This policy aims to prevent algorithmic bias and ensure judicial reasoning remains transparent and personal.
The Details
In a significant move to regulate the use of artificial intelligence in the justice system, Spain’s General Council of the Judiciary (CGPJ) has issued a formal instruction for all judges and magistrates. The directive aims to create a clear and consistent framework for using AI, aligning with emerging national and EU regulations like the AI Act. The primary concern driving this initiative is the potential for generative AI to infringe on fundamental rights and freedoms. This is not about banning technology but about carefully defining its role as a supportive tool, ensuring that ultimate judicial authority remains firmly in human hands.
The new instruction is built on the foundational principle of effective human control. It establishes that AI systems cannot operate autonomously or replace judges in their core duties. Key tenets include the no substitution rule and the principle of judicial responsibility, meaning the judge remains exclusively liable for every aspect of the final ruling, regardless of any AI assistance used. This framework is designed to safeguard judicial independence and prevent scenarios where an algorithm’s inherent biases or errors could improperly influence a case’s outcome without rigorous human oversight.
The guidelines draw a clear line between permitted efficiencies and prohibited delegations of duty. Judges can use approved AI tools for tasks like legal research, summarizing documents, and structuring case information. They may even use AI to generate preliminary drafts of resolutions. However, the instruction explicitly forbids using AI for core judicial functions like evaluating evidence, interpreting the law, or making the final decision. Furthermore, it strictly prohibits using AI for profiling individuals, predicting behavior, or processing specially protected data, providing a critical safeguard for businesses and citizens alike.
Source
Consejo General del Poder Judicial (General Council of the Judiciary), Spain
