Regulating artificial intelligence: Colombia's bid to stay ahead

Colombia is preparing to take a decisive step in regulating artificial intelligence with a bill that seeks to establish a comprehensive framework inspired by international standards such as Europe.

The discussion comes at a time when technology is already a part of daily life and is beginning to transform sectors such as health, education, security, and justice.

In conversation with EL TIEMPO, Mario Castaño, Cintel's technical director, explained the scope of the initiative and the challenges facing the country.

"It's timely that we begin discussing how we can ensure that artificial intelligence truly contributes to the country's growth, ensuring innovation without imposing excessive regulations that ultimately hinder its development," Castaño said.

In his opinion, the Colombian bill follows the logic of the European AI Act , which classifies systems according to their risk level. "The balance between regulation and innovation has always been the focus of discussion in new technologies, and that happy medium is what needs to be found now," he added.

The executive emphasized that the impact of artificial intelligence depends more on the context than the technique used. "An AI that manages the operation of a high-speed train is not the same as one that modifies a photograph," he explained.

Along these lines, he emphasized the need for strong institutions to oversee and audit the use of these tools. He also highlighted regulatory advances in the country, such as the criminalization of the use of deepfakes to cause harm as an aggravating circumstance: "That points the way in the right direction."

Referring to the national capacity to implement the law, he acknowledged that Colombia has talent and a growing ecosystem of research and business, but with infrastructure limitations.

“We don't have data centers with thousands of GPUs to run large-scale models. That remains an investment challenge,” he warned. He emphasized that the foundation of everything remains data management: “Artificial intelligence thrives on data, and without a clear strategy to protect it, any regulatory effort will be incomplete.”

Regarding the possibility of creating their own models, Castaño indicated that there are already projects underway to train for a Colombian LLM, although he warns about the difficulty of maintaining it against global giants. "Creating a model is possible; keeping it alive and competitive is another story. That requires investment, talent, and long-term policies," he said.

One of the most sensitive issues is bias. “Criteria is the main element. Education will be key so that society can identify problematic uses and not simply consume whatever appears on the screen,” he said.

For Castaño, this will be one of the great challenges of the coming years: training citizens with critical thinking skills in the face of what machines produce.

In the private sector, he added, the adoption of these technologies is not advancing at the same pace. "There are companies that say, 'Let's wait a bit to see how things clear up,' and others that take risks and take the plunge. This disparity makes it even more urgent to have local capabilities that offer security and confidence," he noted.

The bill, he concluded, opens a long-term conversation. “We are at the right time to do so. The key is that the regulation protects rights, defines responsibilities, and empowers local talent, without slowing the development of a technology that is already setting the course for the digital world,” he stated.