Colombia - AI Responsible Use Disclosure (91/2023)

Overview

Proyecto de Ley 091 de 2023 (Senate) — "Mediante la cual se establece el deber de información para el uso responsable de la Inteligencia Artificial en Colombia y se dictan otras disposiciones" — was filed on 9 August 2023 and is framed as a transparency-first measure that targets generative AI. The bill mandates that natural and legal persons using generative AI in economic or academic activities inform when AI is used, identify the system used and specify its purpose. The proposal delegates to national ministries the creation of an ethical framework and calls for ongoing review and international cooperation. The official Senate project page and the radicado text provide the primary record for the bill and its procedural steps; see the project entry on the Senate site at Senate — PL 091-23 project page and the deposited text at PL 091-23 text (radicado PDF).

Definitions

The bill contains concise definitional provisions designed to delimit its coverage. Key definitions include: "Artificial Intelligence" as technologies capable of replicating processes of human thought via algorithms; "Generative Artificial Intelligence" as systems able to produce content (text, images, audio, synthetic data); "Duty of Information" as the obligation to disclose use of AI and to specify which AI and for what purpose; and standard administrative definitions used for natural and legal persons. The definitions orient the bill toward disclosure obligations for generative AI specifically, rather than all algorithmic systems.

Governance and Institutional Framework

Proyecto 091/23 assigns design and coordination tasks to national authorities. The bill instructs the Executive—principally the Ministry of Science, Technology and Innovation (MinCiencias) and the Ministry of Information Technologies and Communications (MinTIC)—to design a transversal ethical framework for generative AI and to coordinate updates with other relevant ministries and academic bodies. The text envisions a short initial design period (the radicado text indicates a target of approximately six months for preparing the framework) and periodic reviews. Regulatory enforcement is expected to leverage existing administrative institutions rather than creating a new agency. For primary documentation and the bill record, consult the Senate project entry at Senate — PL 091-23 project page and the text at PL 091-23 text (radicado PDF).

Key Focus Areas

The principal thrust of the bill is on transparency and user awareness. Obligations crafted by the bill include: (1) clear, visible disclosure that AI was used in the generation or mediation of content or decision-making; (2) identification of the AI system or service employed (where reasonably possible); (3) specification of the purpose for which AI was used (e.g., content generation, automated decision support, data augmentation); (4) accessibility requirements so information is identifiable and comprehensible by affected persons; (5) an executive-level ethical framework to interpret and operationalize disclosure duties; and (6) promotion of international cooperation and alignment with global standards. The bill does not establish an EU-style risk classification, mandatory pre-market conformity assessments, or an AI registry in its core text; instead it leans on targeted disclosure as the principal control mechanism while asking executive agencies to build supporting guidance and tools.

Implementation Framework

Operationalization is delegated to the relevant ministries and competent authorities. The Ministries are asked to prepare practical guidance, templates and technical norms to ensure disclosures are standardized and accessible. The text contemplates short deadlines for the initial design of the ethical framework and periodic reviews thereafter, implying the need for inter-agency coordination teams and stakeholder consultation processes (industry, academia, civil society). Implementation tools expected under the bill include public guidance documents, awareness campaigns, and (potentially) non-binding technical standards; where enforcement or sanctions are needed, the bill signals reliance on existing administrative sanctioning regimes rather than creating a bespoke enforcement apparatus.

Monitoring and Evaluation

Monitoring is intended to be carried out by the agencies designated in the implementation phase. The bill specifies periodic review of the transversal ethical framework and expects ministries to evaluate the effectiveness of disclosure requirements, to monitor compliance rates, and to report periodically to the Congress and the public. Given the bill's reliance on existing regulators, monitoring metrics would typically include number of disclosures, accessibility of information, stakeholder feedback, and the issuance of guidance instruments by executive agencies.

Penalties, Liability, and Appeals

The radicado text of Proyecto 091/23 focuses on duties of information and the design of coordinating frameworks and does not set out a detailed, novel punitive regime within its core articles. Instead, enforcement and sanctioning would rely on the powers of competent administrative authorities and on existing legal avenues (e.g., consumer protection and data protection frameworks) unless secondary regulation establishes specific penalties. The official record indicates that, where sanctions are required, implementing authorities would apply applicable administrative sanctions under existing law and offer the ordinary administrative and judicial appeal routes available under Colombian law.

Relationship to Other Instruments

Proyecto 091/23 is one of multiple concurrent Colombian legislative initiatives addressing AI (for example PL-059/23, PL-130/23, PL-200/23). It complements broader policy and legislative projects by focusing specifically on disclosure and transparency rather than seeking a comprehensive regulatory architecture. The bill references alignment with existing data protection rules and invites cooperation with national institutions; it also anticipates that implementing guidance will be coordinated with other pending AI legislative efforts to avoid duplication.

International Alignment

The proposal explicitly intends to align Colombia's approach to AI transparency with international standards and to encourage cooperation with foreign public and private actors. The bill's emphasis on disclosure mirrors global trends toward transparency obligations (for example, aspects of the European Union's AI Act and international ethical recommendations). The project signals Colombia's intent to adopt internationally-informed good practices and to participate in multilateral technical cooperation and capacity-building initiatives.

Implementation Timeline

Sources and References