World Watch/Colombia/Artificial Intelligence

Artificial Intelligence · Colombia

AI regulation in Colombia (2026)

ProposedProyecto de Ley 043 de 2025 (pending Senate approval); CONPES 4144 (National AI Policy, February 2025); Joint Directive 007 (2025) on algorithmic transparency; Law 2502 of 2025 (AI-aggravated identity fraud)Country index 75 · B+

Colombia shaded by its artificial intelligence status

Colombia does not yet have a comprehensive, enacted AI law. The government submitted Proyecto de Ley 043 de 2025 to the Senate in July 2025 — a risk-based framework modelled on the EU AI Act — which remained in legislative debate as of early 2026. In parallel, the country operates under CONPES 4144 (February 2025), a binding national AI policy through 2030, Joint Directive 007 (2025) requiring algorithmic transparency in the public sector, and Law 2502 of 2025 which introduced AI-related aggravating factors in criminal identity fraud.

Key points

Pending Comprehensive AI Bill (PL 043/2025)

Proyecto de Ley 043 de 2025, submitted to the Senate on 28 July 2025 by MinCiencias, proposes a four-tier risk classification (critical, high, limited, minimal) modelled on the EU AI Act, designates MinCiencias as the national supervisory authority, and creates a National AI Observatory. As of early 2026 it had not been enacted.

CONPES 4144 — National AI Policy (2025)

Approved on 14 February 2025 by the National Council for Economic and Social Policy, CONPES 4144 sets the strategic roadmap through 2030 across six axes (ethics & governance, data & infrastructure, R&D, talent, risk mitigation, adoption), mandating 106 actions and approx. COP 479 billion in investment. Led by MinTIC, MinCiencias, DAPRE, and DNP.

Joint Directive 007 — Public Sector Algorithmic Transparency

Issued in 2025 by the Procurador General and the Defensoría del Pueblo (inspired by Constitutional Court Ruling T-067 of 2025), Joint Directive 007 mandates public entities and private actors exercising public functions to disclose design, purpose, and effects of AI/automated decision-making systems they use, and to provide citizen objection channels.

Law 2502 of 2025 — AI-Aggravated Identity Fraud

The first enacted AI-specific law in Colombia amends Article 296 of the Criminal Code to treat AI-assisted identity fraud (including deepfakes) as an aggravating circumstance, increasing applicable fines by up to one-third and introducing Colombia's first statutory definition of 'deepfake'.

EU AI Act Alignment and Risk Classification

PL 043/2025 explicitly adopts the EU AI Act's risk-based approach: critical-risk systems (capable of substantially violating fundamental rights) face prior review and potential prohibition; high-risk systems (education, employment, public services, justice, law enforcement) face enhanced conformity obligations. This signals intent to align with international standards.

Fragmented Legislative Landscape

Beyond PL 043/2025, over 20 separate AI-related bills were before Colombia's Congress as of 2025–2026, covering topics from AI in education to labour rights, making the regulatory landscape fragmented and the final form of any enacted comprehensive law uncertain.

Colombia - other topics

Last verified 5/24/2026 · Orientation, not legal advice - verify against the primary sources linked above. Explore the full world map →