This week, a tense debate over the 2027 Budget unfolded in Congress, and the opposition —and to a lesser extent the ruling coalition— submitted a wave of amendments. After the budget cuts, the opposition alliance parties made good on their warnings and, nearly as a bloc, thousands of amendments to the legislation were filed. Now, the bill faces its passage through the Senate. The observations were so numerous that they broke a record: more than 3,800 amendments. A similar scenario occurred during the passage of the Reconstruction Law, but it was nowhere near the current figure, as some 1,500 amendments were filed at that time. Thus, the 2027 Budget Law became the proposal with the most observations. Finance Minister Jorge Quiroz specified that if the notes from the Mixed Committee are counted, the figure would rise to nearly six thousand. According to the Undersecretary of the Government Secretariat, Gonzalo Arenas, the massive amendments caused surprise in the Executive Branch. "My understanding is that never before have so many amendments been submitted. It is a prerogative that legislators have, and what the Executive Branch must do is take the general ideas and try to deliver solutions or reach reasonable agreements with the opposition," Arenas added. Faced with the tsunami of amendments, Senator Gustavo Sanhueza (UDI), a member of the Special Mixed Committee, went to the Senate leadership to raise the alarms and discuss a possible intervention for these cases. As a result, the Vice President of the Upper Chamber, Iván Moreira (UDI), requested legislative statistics: in the 2026 Budget debate there were no more than 1,500 amendments. Moreira argues that the use of artificial intelligence has contributed to this wave of amendments. The ease with which AI performs tasks has opened the door to a debate about its use in public institutions. In the legal profession, it has already become common to ask a chatbot for certain documents or procedures; however, its use has also led to a controversy in the Judiciary. The signals about AI use were evident in the Santiago Court, where a clerk left an instruction for the AI in a ruling on a civil case, which is strictly prohibited, since the Code of Ethics establishes that no automated system may substitute a judge's decision. In Congress, the debate also opens other angles, such as the obstruction of legislative work due to the large number of amendments, which end up being useless and cause a waste of human resources in their lengthy reviews. Moreira says that this tool is being used indiscriminately. In the corridors of the legislative seat, there is also talk about the use of AI in other areas: speeches that sound robotic or certain verbal tics that betray the footprint of an AI amid a legislator's words. But what is really under discussion are the bills and amendments made with AI. NOVEMBER 3, 2026
FINANCE MINISTER JORGE QUIROZ AT THE SPECIAL MIXED BUDGET COMMITTEE SESSION IN THE SENATE.
PHOTO: DEDVI MISSENE Dedvi Missene
For that reason, in August the Senate already asked the Comptroller's Office for a regulation on the use of AI, although there is still no clarity on possible sanctions or prohibitions; rather, it is a general framework. Secretary General Luis Rojas, who took office in June, translated this concern into a project handled by the IT Department. The engineers of the division created the SenIA application, which is still in the training phase. The initiative must be approved, but the secretary anticipates that it is a system that was fed only with information from Congress and records from the Constitutional Tribunal, as well as other parliamentary interventions and rulings on the admissibility of amendments. Rojas explains that "it will be able to help draft bills, in the sense of gathering the necessary information. For example, if a senator wants to introduce a bill on a certain subject, this system will help relate it to other legislation, votes, and all the necessary background that exists in our records." However, that is not the central point. The application will also be able to deal with massive amendments. The secretary explains that "from the staff's point of view, it becomes harder —and madness— to review more than 3,800 amendments in such a short time. Staff members double their hours to keep up. We need to have this help to make the process faster."
2,242 amendments with AI
In the Chamber of Deputies, the incorporation of AI is more advanced. Secretary General Miguel Landeros says that since 2023 he knew he would have to face this technological advance. He did not want to be left behind. That concern materialized last September with the drafting of a transparency protocol regarding the use of these systems. The idea is for legislators to identify the degree of AI they used in their documents: green for a text without intervention; yellow for partial use; red applies to texts mostly produced with AI, and blue in the case of texts with no human contribution. Of the amendments submitted for the 2027 Budget by deputies, the majority were made with an AI tool. Specifically, 2,242 were validated with a degree of AI; while most carry the yellow label, there is also a group identified with the degree of total AI intervention.
Lawyer Gonzalo Álvarez, director of the Legal Tech program at Universidad Central, focused on the application of AI, highlights that the technology can bring great benefits to Congress. However, he is critical of indiscriminate use. For the academic, the AI identification grade is not enough; legislators must use the tool trained and consciously. "In these cases it is complex, because it is an amendment that can affect millions of citizens: the AI generates a text, but the legislators do not know how or where to provide traceability or explain the result they obtained. One thing is using it to gather information and another is having it do all the work. Moreover, AI has somewhat blurred immediacy: it generates hundreds of options in minutes that collapse legislative work. All these massive amendments do is create a bottleneck and digital garbage. And in this case, it also generates a waste of human time."
Along the same lines, Senate President Paulina Núñez highlights that she is an intensive user of AI, though she makes clear that every request has human supervision. That is why she argues that rather than punishing its use, there must be "criteria and supervised use by each legislator." And she adds: "What obstructs legislative work is not artificial intelligence, but volume without criteria. An amendment is inadmissible whether written by a person or a machine. AI does not submit amendments; legislators submit them: whoever signs is accountable for the content, for its constitutionality, and for the time it makes Congress spend. That said, it would be naive to pretend this is not happening. Today, drafting a hundred amendments takes the time it used to take to draft one, so effort can no longer be the filter."
OCTOBER 21, 2026
PRESIDENT OF THE SENATE, PAULINA NÚÑEZ.
PHOTO: DEDVI MISSENE Dedvi Missene
Precisely to cope with the growing increase in amendments, the Chamber of Deputies is preparing an intelligence system that, if approved, would be operating in March of next year. The purpose is to fight AI with another AI. These are Caminar and Camlex (Chamber of Deputies Expert Legislator), systems created by IT staff of the Chamber that store all the information from previous bills, votes, speeches, and human criteria applied in the admission of amendments and bills. The AI uses its own database and operates with an agentic intelligence model, meaning it can develop decisions autonomously. The rollout also contemplates it being a tool for legislators to access historical records of Congress, thus facilitating the legislative task of drafting bills. The initiative must go through the Science and Internal Regime Committee, which is in charge of approving the resources. The model they are working on incorporates an AI detector. The idea of making chatbot use transparent also escalated to the Executive's messages, although the initiative remained only in informal conversations with the Segpres Undersecretariat. Secretary Landeros says that "with these advances, the Chamber of Deputies becomes a pioneer worldwide. The goal of this is transparency and responsible use; we don't want what has happened in the Judiciary to happen to us, where lawyers use AI and then end up with nobody knowing what was made with AI."
Some legislative seats around the world have already modernized the application of these technologies, as is the case in Brazil, where Congress uses a system to connect with citizens' proposals and, in turn, legislators have the Ulysses AI, which helps staff and advisors. Meanwhile, the European Union blocked all AI applications and tools on the mobile devices of MEPs and their staff for security reasons. In contrast, in the United States, the ChatGPT application is the one most widely used, and they also created an AI tool to verify changes in laws.
For the UCEN academic expert in AI, the issue of security is a key point to consider: "We know that legislators use AI when they declare it with this traffic-light system, but we don't know how they use it, or what documents they feed it to generate information. There can be security risks there if it is not properly regulated."
While both Chambers work on systems to combat the tsunami of amendments, staff members are doubling their hours to review the amendments made with AI by legislators. In the words of the Senate secretary, "it becomes impossible if we continue at this pace; probably next term we will have the application to help us review the admissibility of the thousands of amendments if this volume keeps coming in."