][]

Article

Towards a National Artiﬁcial Intelligence Policy in

Colombia: A Comparative Analysis of International

Frameworks

Hacia una política nacional de inteligencia artiﬁcial en

Colombia: un análisis comparativo de marcos

internacionales

Isabel Carrillo-Gómez ¹, José Mejía-Caballero¹, Sergio Vélez-Muñoz¹

and Andrés

Solano-Barliza¹∗

1

Faculty of Engineering, Universidad de La Guajira, Riohacha, 440001, La Guajira; icarrillog@uniguajira.edu.co;

jmejia@uniguajira.edu.co; svelez@uniguajira.edu.co; andresolano@uniguajira.edu.co

∗

Correspondence: andresolano@uniguajira.edu.co

Citation: Carrillo, I.; Mejía, J.; Vélez, S.; Solano, A. . Towards a National Artiﬁcial Intelligence Policy in Colombia: A Comparative

Analysis of International Frameworks. OnBoard Knowledge Journal 2025, 1, 2. https://doi.org/10.70554/OBJK2025.v01n01.02

Received: 05/09/2025, Accepted: 29/09/2025, Published: 20/10/2025

DOI: https://doi.org/10.70554/OBJK2025.v01n01.02

Abstract: The debate surrounding Artiﬁcial Intelligence (AI) no longer centers on if it should be regulated, but rather on

what that regulation entails and what its orientation should be. While countries across Europe, Asia, and the Americas

have already made progress in AI regulation, Colombia presents a fragmented landscape, featuring 6 withdrawn or

archived bills, 7 active projects, 3 speciﬁc regulatory proposals, and 2 current regulations, all without a clearly deﬁned

path.The methodology used in this analysis is exploratory and qualitative, based on the search, review, and analysis of

scientiﬁc literature and AI regulatory policies. The analysis reveals a stark divergence in regulatory philosophy: one

model, the EU’s guarantee-based approach, prioritizes control, traceability, and protection, establishing a high compliance

risk, while the other, the Asia-Paciﬁc (APAC) model, leans toward ﬂexibility and the active promotion of innovation.

For Colombia, this juncture represents a decisive opportunity to establish clear guidelines that foster innovation and

investment, thereby avoiding rigid regulatory barriers that could hinder market development. The regulation must be

geared toward an ethical, responsible, and competitive framework that drives sustainable sector growth. It is proposed

that Colombian policy adopt a hybrid model, integrating the EU’s protection principles with pro-innovation mechanisms

and a unique component of social justice and labor retraining, positioning the country as a regional benchmark for

responsible regulation adapted to its regional context.

Keywords: Artiﬁcial Intelligence; Ethics; Innovation; Policy; Regulation.

Resumen: El debate sobre la inteligencia artiﬁcial-IA no se centra en si debe regularse, sino en qué implica dicha

regulación y cuál debe ser su orientación. Mientras países de Europa, Asia, y América ya han avanzado en la regulación

de la IA, Colombia presenta un panorama fragmentado, el que cuenta con 6 proyectos de ley retirados o archivados, 7

proyectos activos, 3 proyectos normativos especíﬁcos y 2 normativas vigentes, sin una ruta clara deﬁnida. La metodología

OnBoard Knowledge Journal 2025, 1, 2.

https://revistasescuelanaval.com/obk/

Licensed by Escuela Naval de Cadetes "Almirante Padilla", COL.

This article is freely accessible and distributed under the terms and conditions

of Creative Commons Attribution (https://creativecommons.org/licenses/by/4.0/).

OnBoard Knowledge Journal 2025, 1, 2

2 of 14

utilizada en este análisis es exploratoria y cualitativa, basada en la búsqueda, revisión y análisis de documentos cientíﬁcos

y políticas regulatorias de IA. El análisis muestra dentro del marco de la regulación un modelo garantista de la UE prioriza

el control, la trazabilidad y la protección, estableciendo un alto riesgo de cumplimento y otro modelo de Asía-Paciﬁco

que se inclina por la ﬂexibilidad y el fomento activo de la innovación. Para Colombia, esta coyuntura representa una

oportunidad decisiva para establecer directrices claras que fomenten la innovación y la inversión, evitando barreras

normativas rígidas que diﬁculten el desarrollo del mercado. Así, la regulación debe orientarse hacia un marco ético,

responsable y competitivo que impulse un crecimiento sostenible. Se propone que la política colombiana adopte un

modelo híbrido, que integre los principios de protección de la UE con los mecanismos pro-innovación y un componente

único de justicia social y reconversión laboral, posicionando al país como un referente en la regulación responsable y

adaptada al contexto regional.

Palabras clave: Ética; Innovación; Inteligencia Artiﬁcial; Política; Regulación

1. Introduction

The Organization for Economic Co-operation and Development (OECD) proposes a deﬁnition of

artiﬁcial intelligence (AI) as a “computational system that, for a given set of human deﬁned objectives,

can make predictions and recommendations or take decisions that inﬂuence real or virtual environments.

AI systems are designed to operate with varying levels of autonomy” [10;20]. Along these lines, AI as an

interdisciplinary science focuses on the development of systems capable of performing human cognitive

functions such as learning, reasoning, perception, adaptation, and autonomous decision-making, based on

input data, as well as generating predictions, recommendations, or decisions that impact physical or virtual

environments in order to achieve implicit or explicit objectives [12;22].

In this sense, AI has evolved from being an experimental discipline to becoming a cross-cutting driver

of economic transformation, directly inﬂuencing productivity, operational efﬁciency, and innovation across

numerous sectors. This shift implies that AI has moved from being an emerging technology to a convergent

one, driven primarily by the exponential increase in computational capacity, the massive availability of data

(big data), and continuous advances in machine learning algorithms, which enable integration with diverse

ﬁelds of knowledge [3].

Considering this context, AI requires regulation to ensure that its development and deployment remain

fully aligned with the general interests and requirements of societies [16]. Regulation is understood here as

direct or indirect interventions by the State through binding and mandatory rules (laws, decrees, resolutions,

general instructions, etc.). Such intervention must guarantee a strategic and responsible use of AI, prioritizing

respect for human rights and ensuring that this technological advancement is intrinsically at the service of

human well-being [7]. In response to these AI-related challenges, various States worldwide have deﬁned

strategies for its responsible use, seeking to leverage strong technological management to promote socioeco-

nomic development and enhance public satisfaction or well-being, with the aim of establishing ethical and

legal oversight.

The regulation of AI began around 2017, with the ﬁrst Resolution of the European Parliament to discuss

the regulatory challenges of AI and an initial regulatory attempt in Canada. One of the most important

milestones is undoubtedly the European Union’s AI Act (EU AI Act), which received its ﬁrst approval in that

Parliament in June 2023 [6;21]. This regulation, like the General Data Protection Regulation of 2018, could

inﬂuence various legislative frameworks worldwide, including those of Latin American countries [19]. The

United States (U.S.), for its part, has opted for a different approach, using Executive Order 14110 (signed in

October 2023) as its main governance instrument. This order is based on directing existing federal agencies

to establish their own safety and transparency standards (sectoral regulation), prioritizing innovation and

competitiveness over a central and binding legal framework [23].

In Latin America, there has been progress in deﬁning public policies aligned with technological updating

and the digital evolution of nations. According to the Latin American Artiﬁcial Intelligence Index 2025 [5],

OnBoard Knowledge Journal 2025, 1, 2

3 of 14

only El Salvador and Peru have approved AI laws. In Peru’s case, the law has been in force since July 2023,

while in El Salvador it was approved in the ﬁrst half of 2025. These countries have become benchmarks and

encourage AI regulation and discussion across the region, where 11 of the 19 countries currently have at least

one bill on this subject under discussion.

In the case of Colombia, the State has assumed an active role in the promotion, regulation, and oversight

of this technology, advancing initiatives such as a regulatory framework for AI that emphasizes principles

of transparency, privacy, and non-discrimination. These advances reﬂect an institutional commitment to:

(1) sustainable, responsible, and human-centered technological development; (2) strengthening interinstitu-

tional collaboration; (3) establishing a constant dialogue framework among the institutions involved and

international organizations, ensuring alignment of national initiatives with global standards; (4) promoting

the generation and dissemination of high-quality data repositories; and (5) encouraging the creation of

national open data repositories with criteria of representativeness and quality to facilitate the training and

implementation of AI algorithms [12;20].

However, in terms of AI adoption and regulation, a series of challenges have been identiﬁed, highlight-

ing the lack of a consolidated formal regulatory framework whose development, although underway, is

progressing slowly and generating uncertainty. Moreover, limited strategic capacities at the institutional level

are exacerbated by a persistent connectivity gap, given that 36% of households in rural areas lack internet

access, and service quality is low, with 41% having speeds below 10 Mbps [

notable lack of high-quality data, deﬁcient interoperability, insufﬁcient investment, improper use of personal

data in AI, and limited use of AI in social projects [12 24]. Table 1 presents the AI regulatory bills in Colombia

1;2]. Additionally, there is a

;

that have been processed in the Senate and the House of Representatives.

Table 1. Progress of Artiﬁcial Intelligence Regulation in Colombia

Legislative Status

Bill No. / Year

Quantity

Withdrawn

Archived

Bill No. 021 / 2020 (House of Representatives)

1

3

Bill No. 354 / 2021 (House); 253 / 2022 (Senate); 200 / 2023 (Senate)

Bill No. 059 / 2023 (Senate); 091 / 2023 (Senate); 130 / 2023 (Senate); 005 /

2024 (House); 225 / 2024 (Senate); 293 / 2024 (Senate); 154 / 2024 (House)

Statutory Bill No. 111 / 2022 (Senate); Bill No. 156 / 2023 (House); Bill No.

447 / 2024 (House)

Active

7

3

Speciﬁc Regulatory

Bills

Enacted Regulations

Decree 1078 of 2015 and Decree 403 of 2020

2

Total

16

Source: The authors [20].

Table 1 shows Colombia’s fragmented progress in regulatory and legislative matters over the past

ten years, totaling 16 initiatives. Among these, it is noteworthy that there is one withdrawn bill and three

archived bills. In the Congress of the Republic, there are currently seven active bills, three speciﬁc regulatory

initiatives, and two regulations in force, without providing a clear orientation for a national AI policy [14].

At present, the country is awaiting the discussions of the new bill submitted by the Ministry of Science,

Technology and Innovation in July 2025 Bill No. 43, entitled “An Act to regulate Artiﬁcial Intelligence in

Colombia in order to ensure its ethical, responsible, competitive, and innovative development, and to issue

other provisions.

Against this backdrop, this article aims to conduct a comparative analysis of international AI regulatory

frameworks in order to identify reference principles and governance mechanisms that may inform the

ongoing debate surrounding Colombia’s national AI policy. By examining global regulatory models and

regional experiences, the study seeks to contribute to the formulation of an ethical, responsible, and inclusive

AI governance framework adapted to the Colombian context.

The article is organized as follows. Section 2 outlines the main contributions of the study. Section 3

reviews relevant literature and international experiences in AI regulation. Section 4 describes the exploratory

and qualitative methodological approach adopted. Section 5 presents and discusses the results of the

OnBoard Knowledge Journal 2025, 1, 2

4 of 14

comparative analysis of international regulatory frameworks. Finally, Section 6 summarizes the main

ﬁndings and highlights their implications for the development of a national AI policy in Colombia.

2. Contributions

This research presents the following contributions:

i.

A systematic identiﬁcation and examination of the main international regulatory frameworks for

artiﬁcial intelligence is carried out, analyzing their underlying approaches, guiding principles, and

governance mechanisms.

ii.

iii.

A comparative analysis of the current international AI regulatory landscape is performed, identifying

key opportunities and reference mechanisms that may contribute to strengthening Colombia’s national

artiﬁcial intelligence policy.

A set of policy-oriented recommendations is formulated to support the development of a national

AI policy in Colombia, aimed at ensuring ethical, responsible, and inclusive artiﬁcial intelligence

governance.

3. Related Works

This section presents selected related studies associated with the review of AI regulatory policy in

Colombia and other countries.

15] present a methodology based on a qualitative approach that employs comparative analysis, using

[

document review to examine regulations and ethical principles within the regulatory environments of

Colombia and the European Union (EU). The authors propose a structured methodology for the systematic

selection and analysis of policies and academic articles, which identiﬁed similarities, key divergences, and

regulatory gaps related to the implementation of AI in critical areas such as intellectual property, algorithmic

transparency, and data protection. One of the main conclusions emphasizes the urgent need to detect and

mitigate biases in AI systems in order to prevent adverse effects on fundamental rights and prohibited

discrimination, highlighting the importance of audits and preventive measures to ensure inclusive and

equitable development.

Similarly, [11] conducted an independent comparative analysis of national AI strategies in the public

sector of six Latin American countries: Argentina, Brazil, Chile, Colombia, Mexico, and Uruguay. Method-

ologically, they structured their analysis around categories related to objectives, guiding principles, lines

of action, and overall vision/goals, with a clear focus on the ethical and human rights dimensions of these

policies. Through this approach, they identiﬁed strategic characteristics that revealed similarities and differ-

ences among the countries. Their ﬁndings indicate that AI implementation in the public sector is at a stage of

signiﬁcant, albeit still emerging, progress, and that there is convergence in ethical and human rights–based

approaches. The differences identiﬁed are largely determined by the distinct political, administrative, and

technological contexts inherent to each country.

4. Materials and Methods

The methodology employed is exploratory in nature with a qualitative approach [2]. Its design incorpo-

rated a structured process for searching, reviewing, and analyzing scientiﬁc documents and AI regulatory

policies. The steps followed are detailed below (Figure 1).

4.1. Search

In the search process, a systematic selection strategy was adopted. The search process was divided into

two groups of documents for comparative analysis. On the one hand, scientiﬁc literature was reviewed, and

on the other, AI policy documents from different countries in Europe, Asia, and the Americas were selected.

OnBoard Knowledge Journal 2025, 1, 2

5 of 14

Figure 1. Block diagram of the methodological process of the study.

Source: The authors.

4.1.1. Criteria for the Selection of Scientiﬁc Literature

In the article selection process, inclusion and exclusion criteria were established and applied to the

Scopus and ScienceDirect databases. The keywords used in the search strings were in English and included:

“artiﬁcial intelligence,” “public policy,” and “law.”

4.1.2. Criteria for the Inclusion of AI Policy Documents

In the second group of policy documents, primary governmental and international sources were used

for selection, regardless of language, in order to ensure the rigor and ﬁdelity of the comparative analysis.

4.2. Review

The methodological review process was conducted using a qualitative approach structured in two

phases. The ﬁrst phase consisted of identifying and selecting scientiﬁc articles, which were classiﬁed into

three main thematic categories reﬂecting the pre-existing academic discussion: 1) Global governance and

foundational ethical frameworks; 2) Analysis of national and regional policies (Latin America); and 3)

Technological regulation and challenges.

The second phase focused on the identiﬁcation and selection of ofﬁcial AI regulatory policy documents.

In this study, the two most inﬂuential regulatory models were selected, using a comparative approach to

contrast rights-based (guarantor) models (such as that of the European Union) with pro-innovation models

(such as those of the Asia-Paciﬁc—APAC region). This comparison made it possible to identify reference

mechanisms and opportunities to strengthen Colombia’s policy model.

4.3. Data Analysis

The analysis of the reviewed documents, both from academic literature and public policy, focused on a

comparative examination based on two central thematic axes. These guiding axes were oriented toward the

typologies of regulatory models and the reference mechanisms for generating innovation and development

in the regions where they are implemented.

5. Results

This section presents the results and discussion related to the object of study of this article, which

focuses on identifying the main global regulatory frameworks, analyzing the international regulatory context

to determine opportunities and reference mechanisms, and ﬁnally proposing recommendations aimed at

strengthening Colombia’s national AI policy.

5.1. Main International Regulatory Frameworks

The following section presents the results concerning the main international AI regulatory frameworks

and the relevant elements identiﬁed through the literature review.

OnBoard Knowledge Journal 2025, 1, 2

6 of 14

5.1.1. International Regulatory Frameworks for AI

Figure 2 provides a detailed overview of the main AI regulatory frameworks that were formalized

worldwide during 2024. These frameworks demonstrate the active participation and leadership of the regions

of Europe, Asia, and the Americas in shaping AI policy at the global level.

Figure 2. Mentions of AI in legislative procedures by country, 2024.

Source: The authors [25].

Figure 2 shows the intensity with which AI was mentioned in legislative proceedings across 75 ge-

ographic areas during 2024. At the global level, Spain led this activity with 314 mentions, followed by

Ireland (145) and Australia (123). These nations set the global standard, with activity far exceeding that

of the remaining 57 areas analyzed [26]. In Sub-Saharan Africa and other regions with lower levels of

institutional development, the percentages are much lower, indicating that there is still signiﬁcant room for

the development of AI policies. Regarding other regions, the percentage distribution is as follows: North

America at 100%, Europe at 65%, the Middle East and North Africa at 30%, East Asia and the Paciﬁc at 25%,

Latin America and the Caribbean at 19%, and Sub-Saharan Africa at a very low level, with 4%.

Among the advances in regulation, the European Union stands out, having implemented Regulation

(EU) 2024/1689, which is based on a graduated risk approach and seeks to ensure respect for fundamental

rights and the safety of AI systems [9]. Table 2 presents the central axes of its policies and details of each of

these regulatory pillars for AI.

OnBoard Knowledge Journal 2025, 1, 2

7 of 14

Table 2. Core Axes of the European Union Artiﬁcial Intelligence Regulatory Policy

Core Axis

Axis Description

The regulation deﬁnes Unacceptable Risk systems (prohibited), High-Risk, Limited Risk,

and Minimal Risk systems.

Risk Classiﬁcation

Includes systems that pose an unacceptable risk to fundamental rights, such as manip-

ulative AI exploiting vulnerabilities, social scoring, and with very limited exceptions,

real-time remote biometric identiﬁcation in public spaces.

Mandatory requirements such as detailed technical documentation, data quality as-

surance (training, validation, and testing), human oversight, and a Risk Management

System throughout the AI lifecycle.

Non-compliance with data requirements for high-risk systems may result in ﬁnes of

up to €35 million or 7% of global annual turnover (whichever is higher), or violations

related to prohibited practices.

Prohibited Practices

High-Risk Systems

Sanctions

Source: The authors [9;15].

The Asia–Paciﬁc region represents a regulatory mosaic, as it does not have a uniﬁed approach, instead

exhibiting a duality between strict state control and the intensive promotion of innovation. Table 3 presents

the AI regulatory frameworks in the Asia–Paciﬁc region.

Table 3. Artiﬁcial Intelligence Regulatory Frameworks in the Asia-Paciﬁc (APAC) Region

Country

Regulatory Approach

Key Provisions

Mandatory labeling (including watermarking) and strict content

control for Generative AI.

China

Prescriptive and Control-

Oriented

Promotion of innovation; ﬂexible intellectual property approach

for AI model training.

Practical governance frameworks (Model AI Governance Frame-

work) and voluntary certiﬁcation (AI Verify).

Promotion of competitiveness with safeguards; requires prior

veriﬁcation for “High-Impact AI” and transparency obligations.

Japan

Flexible (Soft Law)

Voluntary Governance

Hybrid Model

Singapore

South Korea

Source: The authors [18].

In Latin America, the most active countries were Mexico, Brazil, Colombia, and Peru, which appear in

the medium to high categories of the chart (between 56 and 250 mentions), reﬂecting a signiﬁcant regional

effort to advance AI governance, despite the regulatory gap with global leaders [25]. Figure 3 shows a map

representation of Latin American countries with AI laws and draft bills on AI regulation.

Figure 3 provides an overview of the ongoing legislative activity on artiﬁcial intelligence across Latin

America. Although data for 2024 indicate a strong regional trend, 11 of the 19 countries analyzed are actively

debating or advancing AI related bills, the formal implementation of regulatory frameworks remains limited.

According to the Latin American Artiﬁcial Intelligence Index 2025 [5], the current regulatory leaders in the

region are Peru and El Salvador, the only two countries that have formally enacted speciﬁc AI legislation.

Peru set an early precedent, with its law entering into force in July 2023. Subsequently, El Salvador joined

this vanguard, securing approval of its law during the ﬁrst half of 2025.

Likewise, according to the Latin American Artiﬁcial Intelligence Index [5], only six of the 19 countries

that make up Latin America Venezuela, Cuba, Guatemala, Honduras, Jamaica, Uruguay, and Bolivia do

not have initiatives in this area. This situation highlights the need for these governments to design policies

for AI regulation in order to facilitate digital transformation and address the various ethical and regulatory

challenges that arise with the penetration of this technology into society. Within this framework, Colombia

is consistently identiﬁed as a nation with an advanced regulatory process in AI compared to its regional

counterparts. Nevertheless, this advanced position is accompanied by the urgent need to accelerate and

deepen the internal legislative debate, which is essential to translate the country’s current regulatory proposals

OnBoard Knowledge Journal 2025, 1, 2

8 of 14

Figure 3. Latin American countries with draft laws on AI.

Source: The authors [5].

into effective and robust legal frameworks aligned with the speciﬁc requirements of AI governance in the

Colombian context.

5.1.2. Review of the Literature on AI Regulation

The literature review of the articles selected from the databases made it possible to identify and classify

thematic categories, the results of which are presented in Table 4. Table 4 synthesizes the main thematic axes

that stand out in AI regulation: (1) global governance and foundational ethical frameworks; (2) analysis of

national and regional policies (Latin America); and (3) technological regulation and challenges.

5.2. Identifying Opportunities and Reference Mechanisms to Strengthen AI Policy in Colombia

The comparative analysis of international regulatory frameworks reveals a fundamental tension between

fostering innovation and protecting rights (control), identifying various global reference mechanisms that

can strengthen AI policy in Colombia.

5.3. Safeguard-Based Model and EU Compliance Reference Mechanisms

The EU model is structured around a tiered risk-based framework, which deﬁnitively establishes the

standard of safeguards for regulation. Consequently, its governance focuses largely on traceability, risk

management throughout the product life cycle, and strong sanctioning powers (with ﬁnes that may reach up

to 7% of a company’s annual global turnover). This rigor sets a precedent for the high risk of non-compliance

and is applied extraterritorially, ensuring comprehensive protection for European users even when the

provider is based outside the EU.

Within this European context, Italy has been a pioneer in implementing the EU AI Act (Regulation

2024/1689) through its own legislation (DDL), placing particular emphasis on an “anthropocentric” concept

[8]. This approach clearly contrasts with Spain’s draft bill, which adopts a strongly punitive and safeguard-

based spirit, allocating more than half of its regulatory proposal to sanctions and infringements. Speciﬁcally,

while Spain generally focuses on sanctions, Italian law establishes prison sentences (from 1 to 5 years) for the

malicious use of deepfakes that cause harm (for example, impersonating public ﬁgures) and, more broadly,

OnBoard Knowledge Journal 2025, 1, 2

9 of 14

Table 4. Categories of Analysis of the Literature on AI Policy

Prior Literature Cate- Main Focus of Previous Similarities

Differences

gory

Global Governance Focus on establishing ethical Recognition that regulation The theoretical framework

and Foundational principles and the need for is mandatory and neces- integrates innovation, ethics,

Ethical Frameworks regulation to ensure ethical, sary to protect human rights and

[27] transparent, and responsible and manage systemic risks. proposing

Studies

digital

sovereignty,

typology of

a

AI development. Address Adoption of AI conceptual regulatory models (e.g., EU

the profound impact of AI on frameworks [3].

fundamental rights and pro-

vs. APAC), moving beyond

isolated ethical frameworks.

ductive systems [19].

National and Re- Focus on the design of na- Shared interest in digital Unlike regionally focused

gional Policy Analy- tional strategies and the anal- modernization and public studies, this article situates

sis (Latin America) ysis of ethical and human policy formulation in Latin the Colombian case within

[13;21]

rights approaches in speciﬁc America. Agreement on the a renewed global perspec-

Latin American countries.

need to incorporate rights tive by comparing it with

protection and mitigate algo- key models such as the EU,

rithmic bias.

China, and other APAC coun-

tries.

Technological Reg- Address the need for regula- Highlight the need for regu- The current analysis empha-

ulation and Chal- tory frameworks in the face latory frameworks as institu- sizes that the debate has

lenges [4]

of uncertainty and the rapid tional responses that recon- tended to focus on a market-

evolution of converging tech- cile innovation with the pro- driven perspective, prioritiz-

nologies.

tection of human and demo- ing industry interests, a chal-

cratic values.

lenge that requires going be-

yond minimum regulatory

frameworks.

Source: The authors.

increases penalties for fraud. This divergence highlights a key philosophical debate regarding whether

national AI regulation should prioritize a human-centered development vision (Italy) or a law-enforcement

approach predominantly focused on sanctions (Spain) [17].

5.4. Lessons on Implementation and Flexibility (APAC)

In contrast, the APAC landscape offers valuable lessons on implementation and ﬂexibility. South Korea’s

approach adopts an intermediate position that imposes clear obligations for “high-impact” AI systems and

requires the designation of a national agent for foreign providers, thereby establishing targeted oversight. At

the ﬂexibility end of the spectrum, Japan and Singapore actively seek to avoid regulatory barriers through

the use of soft law and the implementation of regulatory sandboxes, promoting innovation. China, for its

part, imposes prescriptive control focused on the security of generated content, requiring technical solutions

such as watermarking. The reference mechanisms identiﬁed in APAC demonstrate that it is possible to foster

innovation (Japan/Singapore) or impose speciﬁc control (South Korea/China) [27].

By reviewing the typologies of regulatory models (EU vs. APAC) and situating the Colombian case

within a renewed global perspective, this comparative exercise makes it possible to identify Colombia’s

regulatory proposal as a hybrid framework. It adopts best practices from the EU’s risk-based approach

and its ethical principles, while integrating innovation-promoting practices (sandboxes) from the APAC

model. In addition, it is distinguished by introducing a strong component of social and labor justice

(reskilling and just transition) and by addressing the criminal implications of the malicious use of AI (such as

deepfakes). The analysis highlights that the global regulatory debate has tended to focus on a market-oriented

perspective, a challenge that requires Colombian policy to go beyond minimum frameworks. The country’s

great opportunity lies in using this hybrid vision to reconcile innovation with the protection of human

OnBoard Knowledge Journal 2025, 1, 2

10 of 14

and democratic values, strengthening digital sovereignty, human talent, and the social appropriation of

technology.

5.5. Recommendations for Strengthening the National AI Policy under Consideration in Colombia

The analysis of global regulatory models is applied directly to the Colombian bill, identifying critical

areas for its strengthening and effective implementation. The regulatory framework currently being devel-

oped in Colombia is an ambitious bill that seeks comprehensive regulation aligned with the principles of

key international organizations (UNESCO, OECD, EU). The central elements of the policy currently under

consideration are presented in Table 5.

Table 5. Core pillars of the EU AI regulatory policy

Core Pillar

Pillar Details

The Act deﬁnes systems as Unacceptable Risk (prohibited), High Risk, Limited Risk,

and Minimal Risk.

Risk Classiﬁcation

Includes systems that pose an unacceptable risk to human rights, such as

manipulative AI exploiting vulnerabilities, social scoring, and under very strict

exceptions, real-time remote biometric identiﬁcation in public spaces.

Mandatory requirements such as detailed technical documentation, data quality

assurance (training, validation, testing), human oversight, and a Risk Management

System throughout the AI lifecycle.

Non-compliance with data requirements for high-risk systems may result in ﬁnes of

up to EUR 35 million or 7% of the company’s global annual turnover (whichever is

higher), or violations related to prohibited practices.

Prohibited Practices

High-Risk Systems

Sanctions

Source: The authors [9;15].

Taking into account the need to reconcile innovation with the protection of rights and the lessons drawn

from international models, Figure 4 proposes a set of recommendations to strengthen AI policy in Colombia:

Figure 4. Recommendations to strengthen AI policy in Colombia.

Source: The authors.

OnBoard Knowledge Journal 2025, 1, 2

11 of 14

Figure 4 presents as its ﬁrst recommendation the structuring of AI governance through a multilevel

model. While MinCiencias should act as the national authority, it is suggested that it be supported by a

Multi-Stakeholder AI Council and a National Observatory on AI and Digital Rights, with active participation

from academia, civil society, territorial entities, and the private sector. This would facilitate more transparent,

democratic regulation that is adaptable to rapid technological change. The next recommendation concerns

establishing rigorous standards for algorithmic transparency and traceability, including explicit regulatory

obligations regarding traceability, explainability, and visible labeling of AI-generated content (watermarking).

This is especially necessary in contexts of automated decision-making that affect rights in sensitive sectors

such as health, justice, education, credit, and public services.

Likewise, Figure 4 recommends that the policy adopt a differential, territorial, and ethical approach,

explicitly incorporating differential, territorial, and ethnic perspectives that recognize the potential effects

of AI in reproducing structural inequalities. This includes: (1) conducting mandatory impact assessments;

(2) guaranteeing prior participation of ethnic communities; and (3) recognizing collective rights over data

generated within their territories. Another recommendation is the establishment of mechanisms for just

transition and ﬁnancing. To this end, it is essential to create a Just Digital Transition Fund to ensure ﬁnancing

for labor reconversion, talent training, and technological appropriation in prioritized regions. This would

ensure that automation and digitalization do not deepen existing social gaps.

Another recommendation, as shown in Figure 4, is the gradation of sanctions according to the severity

of the infringement (use of prohibited systems, non-compliance with high-risk requirements, lack of trans-

parency) and the actual or potential harm to fundamental rights. It is proposed that a signiﬁcant portion of

the ﬁnes collected be allocated to the Just Transition Fund and to ﬁnancing research projects on responsible

AI in public and regional universities.

Finally, the internal discussion on Colombia’s AI regulatory policy should be grounded in an anthro-

pocentric core as the central regulatory premise. In this sense, it is necessary for the regulatory framework

to strategically avoid imposing excessively rigid regulatory barriers that could actively hinder foreign in-

vestment and limit the country’s capacity to catalyze the development of domestic technological products.

The approach should focus on establishing clear rules of participation and well-deﬁned operational limits

for AI deployment. For this reason, the path forward requires maintaining a philosophical deliberation that

positions the country as a signiﬁcant contributor of ethical and technological value. Therefore, this approach

must preserve the necessary levels of policy ﬂexibility to effectively adapt and integrate AI as a fundamental

support mechanism for human development in the national context.

6. Conclusions

The regulation of artiﬁcial intelligence in Colombia represents a decisive strategic opportunity to

establish clear guidelines that foster innovation and attract investment, making it imperative to consciously

avoid excessive regulatory rigidity or the creation of barriers that could inhibit market development; the

resulting framework should be oriented toward cultivating an ethical, responsible, and highly competitive

ecosystem capable of ensuring sustainable sectoral growth and ﬁrmly positioning the national territory as a

preferred destination for technological investment. In the regulatory process currently underway in Colombia,

the primary guideline must be to unequivocally anchor the entire framework to an anthropocentric core,

guaranteeing the preservation of fundamental rights above all else; accordingly, emphasis should be placed

on the institutionalization of agile, multi-actor governance mechanisms that provide the ﬂexibility necessary

for continuous adaptation, enabling the country to design a distinctive and robust model that successfully

reconciles compliance assurance with the strategic imperative of national innovation and inclusion. Finally,

Colombia’s AI policy should be implemented through a multilevel governance model that integrates multiple

stakeholders, ensuring transparency, algorithmic traceability, and explainability, while adopting a differential,

territorial, and ethical approach to mitigate structural inequalities.

OnBoard Knowledge Journal 2025, 1, 2

12 of 14

Author Contributions: Isabel Carrillo-Gómez: Conceptualization, Methodology, Validation, Formal analysis, Inves-

tigation, Resources, Data curation, Writing – original draft, Writing – review & editing. José María Mejía-Caballero:

Conceptualization, Methodology, Formal analysis, Investigation, Resources, Data curation, Writing – original draft,

Writing – review & editing. Sergio Vélez-Muñoz: Conceptualization, Writing – original draft, Writing – review & editing,

Supervision, Project administration. Andrés Solano-Barliza: Conceptualization, Supervision, Project administration,

Funding acquisition.

All authors have read and agreed to the published version of the manuscript. Refer to the taxonomía CRediT for term

explanations. Authorship should be limited to those who have contributed substantially to the work reported.

Funding: This research received no external funding.

Institutional Review Board Statement: Not applicable, since the present study does not involvehuman personnel or

animals.

Informed Consent Statement: This study is limited to the use of technological resources, so nohuman personnel or

animals are involved.

Conﬂicts of Interest: Under the authorship of this research, it is declared that there is no conﬂict of interest with the

present research.

References

1. Arrubla-Hoyos, W., Ojeda-Beltrán, A., Solano-Barliza, A., Rambauth-Ibarra, G., Barrios-Ulloa, A., Cama-Pinto, D.,

Arrabal-Campos, F. M., Martínez-Lao, J. A., Cama-Pinto, A., and Manzano-Agugliaro, F. (2022). Precision agriculture

and sensor systems applications in colombia through 5g networks. Sensors, 22(19).

2. Barrios-Ulloa, A., Solano-Barliza, A., Arrubla-Hoyos, W., Ojeda-Beltrán, A., Cama-Pinto, D., Arrabal-Campos, F. M.,

and Cama-Pinto, A. (2025). Agriculture 5.0 in colombia: Opportunities through the emerging 6g network. Sustainability,

17(15):1–28.

3. Brynjolfsson, E. and McAfee, A. (2017). Artiﬁcial intelligence, for real. Harvard Business Review, 1(1):1–31.

4. Bussacarini, M. (2025). Global readiness for cybersecurity and ai: Assessing the landscape and charting the path

forward. In Chakraborty, M., Chakrabarty, S. P., Penteado, A., and Balas, V. E., editors, Proceedings of the 5th International

Ethical Hacking Conference, pages 3–15. Springer Nature Singapore.

5. Comisión Económica para América Latina y el Caribe (CEPAL) (2025). Índice latinoamericano de inteligencia artiﬁcial

(ilia) 2025. Technical Report LC/TS.2025/68, CEPAL.

6. Crum, B. (2025). Brussels effect or experimentalism? the eu ai act and global standard-setting. Internet Policy Review,

14(3):1–21.

7. Du, J. (2025). Toward responsible and beneﬁcial ai: Comparing regulatory and guidance-based approaches. arXiv

preprint.

8. Ekdal, D. (2021). Normative power europe & ai: How the eu intends to normatively govern artiﬁcial intelligence

technologies. Master’s thesis.

9. European Parliament and Council of the European Union (2024). Regulation (eu) 2024/1689 laying down harmonised

rules on artiﬁcial intelligence.

10. Figueroa Mosquera, D. (2024). La implementación de la inteligencia artiﬁcial en el sector público de colombia: una

revisión de literatura. Unpublished manuscript.

11. García Benítez, V. H. and Ruvalcaba-Gómez, E. A. (2021). Análisis de las estrategias nacionales de inteligencia

artiﬁcial en américa latina: estudio de los enfoques de ética y de derechos humanos. Revista de Gestión Pública,

10(1):5–32.

12. Grajales, J. F. V., Galeano, M. S. M., and Ortiz-Clavijo, L. F. (2025). National artiﬁcial intelligence policy. Ratio Juris

(UNAULA), 20(40).

13. Guenduez, A. A. and Mettler, T. (2023). Strategically constructed narratives on artiﬁcial intelligence: What stories are

told in governmental artiﬁcial intelligence policies? Government Information Quarterly, 40(1):101719.

14. Gutiérrez, J. D. (2024). Regulación sobre ia. Foro Administración, Gestión y Política Pública.

15. Jiménez Serranía, V., Calderón Marenco, E. A., Agón López, J. G., and Ravelo Franco, G. (2025). La regulación de la

inteligencia artiﬁcial. DIXI, 27:1–23.

16. Kilic, B. (2025). Ai, innovation and the public good: A new policy playbook.

17. Lagioia, F. and Sartor, G. (2020). Ai systems under criminal law. Philosophy & Technology, 33(3):433–465.

OnBoard Knowledge Journal 2025, 1, 2

13 of 14

18. Lee, S.-H. and Wong, P. S. (2024). Regulatory pathways and challenges in asia-paciﬁc ai policy. Journal of Asian

Technology and Development, 23(2):150–175.

19. León, C. (2023). La carrera por la regulación de la inteligencia artiﬁcial.

20. Ministerio de Ciencia, Tecnología e Innovación (2025). Proyecto de ley 043 de 2025.

21. Radanliev, P. (2025). Frontier ai regulation: What form should it take? Frontiers in Political Science, 7.

22. Russell, S. J. and Norvig, P. (2016). Artiﬁcial Intelligence: A Modern Approach. Pearson.

23. Sloane, M. and Wüllhorst, E. (2025). A systematic review of regulatory strategies and transparency mandates in ai

regulation in europe, the united states, and canada. Data & Policy, 7:e11.

24. Solano-Barliza, A., Valls, A., Acosta-Coll, M., Moreno, A., Escorcia-Gutierrez, J., De-La-Hoz-Franco, E., and Arregoces-

Julio, I. (2024). Enhancing fair tourism opportunities in emerging destinations by means of multi-criteria recommender

systems: The case of restaurants in riohacha, colombia. International Journal of Computational Intelligence Systems, 17(1).

25. Stanford Institute for Human-Centered Artiﬁcial Intelligence (2024a). Ai index report 2024. https://aiindex.stanford.

edu. Accessed 2024.

26. Stanford Institute for Human-Centered Artiﬁcial Intelligence (2024b). Ai index report 2024.

27. Zaidan, E., Truby, J., Ibrahim, I. A., and Hoppe, T. (2025). Hybrid global governance for responsible and inclusive

artiﬁcial intelligence. Technology in Society, page 103159.

Authors’ Biography

Isabel Carrillo-Gómez Professor at the Universidad de la Guajira.

José María Mejía-Caballero Professor at the Universidad de la Guajira.

Sergio Vélez-Muñoz Professor at the Universidad de la Guajira.

Andrés Solano-Barliza Professor at the Universidad de la Guajira.

OnBoard Knowledge Journal 2025, 1, 2

14 of 14

Disclaimer/Editor’s Note: Statements, opinions, and data contained in all publications are solely those of the individual

authors and contributors and not of the OnBoard Knowledge Journal and/or the editor(s), disclaiming any responsibility

for any injury to persons or property resulting from any ideas, methods, instructions, or products referred to in the

content.