Colombia PL 043/2025S bill would name MinCiencias AI authority | TLY

Colombia PL 043/2025S bill would name MinCiencias the national AI authority

By Anthony Guerriero, Editor, AI Regulation News. Bogota, dateline 3 December 2025. Last verified: 2026-07-25.

Most of the attention on Colombian AI policy has gone to the national strategy document, CONPES 4144 of 2025. The harder instrument to read, and the one that would actually create binding institutions if it passes, is a bill: Proyecto de Ley No. 043 de 2025 Senado, paired in the lower house as 324 de 2025 Camara. It is an omnibus draft that would name a single national AI authority, is structured around EU-style risk tiers, and would set up a regulatory sandbox. Because it is still moving through Congress, few English-language trackers have quoted the operative articles. This piece works from the official first-debate report filed with the Sixth Committee, and nothing in it is settled law.

What Proyecto de Ley 043 de 2025 Senado would do

The bill's full title, in the filed text, is "Por medio del cual se regula la inteligencia artificial en Colombia para garantizar su desarrollo etico, responsable, competitivo e innovador, y se dictan otras disposiciones" (roughly, "regulating artificial intelligence in Colombia to guarantee its ethical, responsible, competitive, and innovative development, and issuing other provisions").

Per the report's Tramite section, it was filed on 28 July 2025 by the national government, led by the Minister of Science, Technology and Innovation, Angela Yesenia Olaya Requene, together with the Minister of National Education and the Minister of Information and Communications Technologies (MinTIC), plus a long list of senators and representatives. It was published in Gaceta del Congreso No. 1390 de 2025.

The substance is built around the shape of the EU AI Act. Article 5 would sort AI systems into risk classes. The explanatory report describes four levels: prohibited or unacceptable-risk uses, high-risk systems subject to impact assessments and audits, systems with specific transparency duties toward the user, and minimal-risk or no-risk systems left to self-regulation. The report itself states that this "adopts the standards of the European Artificial Intelligence Regulation (EU, 2024)," which is the report's own framing rather than a clause binding the EU text into Colombian law by reference. Article 6 would add an AI Impact Assessment, and Article 13 would create a regulatory sandbox, a supervised testing environment for high-impact technologies.

The proposed AI authority: MinCiencias in the chair, SIC on enforcement

The institutional core sits in Chapter IV. The report summarizes it directly:

"El Capitulo IV crea la Autoridad Nacional para la Inteligencia Artificial, en cabeza del Ministerio de Ciencia, Tecnologia e Innovacion (MinCiencias), encargada de orientar la implementacion de la ley, coordinar la gobernanza y emitir conceptos tecnicos vinculantes en materia de riesgo."

In English, the report says Chapter IV would create the National Authority for Artificial Intelligence, headed by MinCiencias, charged with steering the law's implementation, coordinating governance, and issuing binding technical opinions on risk. Article 7 puts the assignment in statutory language:

"El Ministerio de Ciencia, Tecnologia e Innovacion sera la autoridad nacional competente para orientar la implementacion de esta ley y coordinar la gobernanza de la IA en Colombia."

That reads, in English, as: the Ministry of Science, Technology and Innovation shall be the national competent authority to steer the implementation of this law and coordinate AI governance in Colombia. The ministry's role, as drafted, would be technical, advisory, and coordinating. It would be able to adopt guidelines, technical standards, and good practices, and for high-risk or prohibited systems it would be able to issue technical recommendations or set additional requirements. The teeth are in one clause: the authority would be empowered to issue binding technical opinions on the risk level of AI systems, which would then feed the decisions of bodies that hold inspection, oversight, or control powers.

Enforcement muscle would be placed elsewhere. Chapter IX would designate the Superintendencia de Industria y Comercio (SIC) as the inspection and oversight authority for data protection, free competition, and consumer rights, and the bill contemplates fines, temporary suspension, or shutdown of AI systems in cases of serious non-compliance. Two supporting bodies would round out the design: a cross-sector Coordination Committee and a National Advisory Council of AI Experts. Article 8 would tie the whole scheme back to CONPES 4144 de 2025, so the bill is meant to give the strategy a statutory backbone rather than replace it.

Status: a draft in tramite, moving under a message of urgency

This is not law. It is a bill in active passage. The timeline in the official report is specific: filed 28 July 2025; sent to the Senate's Sixth Constitutional Committee on 14 August 2025; Senator Pedro Hernando Florez Porras named as first-debate rapporteur on 2 September 2025; a request for a message of urgency filed on 8 September 2025 to speed the joint-chamber track; and, on 18 November 2025, Representative Gerson Lisimaco Montano Arizala named as sole rapporteur in the Camara. On 3 December 2025 the rapporteurs filed a positive ponencia for the first debate. The government's own announcement confirmed the milestone:

"El Gobierno Nacional celebro la radicacion de la ponencia positiva para el primer debate del Proyecto de Ley 043 de 2025 Senado - 324 de 2025 Camara, iniciativa que regula el uso, desarrollo e implementacion de la Inteligencia Artificial en el pais."

Translated: the national government welcomed the filing of the positive report for the first debate of the bill, which regulates the use, development, and implementation of AI in the country. A positive report clears the way for a first-debate vote in committee. It does not enact anything. Because a message of urgency is in play, the two chambers can debate jointly, which can shorten the road, but the text can still change at every stage.

How it compares with other LatAm AI-framework bills

Colombia is one of several countries in the region trying to legislate a horizontal AI framework. The table sets the three most-cited bills side by side. All three were still drafts at the time of writing.

FeatureColombia PL 043/2025SBrazil PL 2338/2023Chile (AI framework bill)
Named lead authorityWould name MinCiencias as national AI authority; SIC for oversightCoordinating role contemplated for the data protection authority (ANPD)Lead role contemplated for the data protection agency once created
Risk modelFour tiers, structured around the EU AI ActRisk-based, excessive-risk and high-risk categoriesRisk-based, aligned to EU-style tiers
OriginGovernment bill, multi-ministrySenate committee substitute textExecutive-backed bill
StatusDraft, first debate pending (Dec 2025)Draft, advanced in SenateDraft, in committee
SandboxWould create one, Article 13Regulatory experimentation provisionsProvisions under discussion

The point of difference worth citing is the authority choice. Brazil and Chile lean toward a data-protection regulator as the natural home for AI oversight. Colombia's bill would instead put the science and technology ministry in the coordinating chair and split enforcement out to the SIC. That is a distinct institutional bet, and it is the detail most likely to be quoted as the bill advances.

What the bill does NOT do

Key Facts

Instrument
Proyecto de Ley No. 043 de 2025 Senado / 324 de 2025 Camara (draft AI framework bill). Source: Camara de Representantes first-debate report.
Issuer / sponsors
National government: MinCiencias (Minister Angela Yesenia Olaya Requene), Ministry of National Education, MinTIC, plus congressional co-sponsors. Filed 28 July 2025 per the report; published in Gaceta del Congreso No. 1390 de 2025.
Proposed AI authority
Article 7 would name MinCiencias as the national competent authority for AI governance; the Superintendencia de Industria y Comercio would handle inspection and oversight. Neither role exists yet.
Who would be covered
Developers and deployers of AI systems in Colombia, classified across four risk tiers under Article 5, if the bill is enacted.
Penalty / consequence
Would authorize fines, temporary suspension, or shutdown of AI systems for serious non-compliance (Chapter IX); binding technical risk opinions from the authority. Technical rules would be issued within 24 months of promulgation.
Status
Draft, en tramite. Positive first-debate ponencia filed 3 December 2025; first debate pending in the Sixth Constitutional Committee. No legal force.

FAQ

Is PL 043 de 2025 a law in Colombia? No. It is a draft bill still in Congress. As of the positive first-debate report of 3 December 2025 it was awaiting its first committee debate and would have no legal force unless enacted.

Which body would be Colombia's AI authority under the bill? Article 7 would name MinCiencias as the national competent authority for AI governance, with the Superintendencia de Industria y Comercio handling inspection and oversight for data protection, competition, and consumer rights.

Does the draft bill copy the EU AI Act? It is structured around EU-style risk tiers. The explanatory report says the four-tier classification adopts the standards of the European AI Regulation of 2024, which is the report's own characterization rather than a binding incorporation by reference.

Primary sources and related coverage