Wanted! Someone to lead and coordinate AI regulation
Alfredo González Briseño
10/12/20269 min read
As conversations on the risks from AI grow—including how it could wipe out humans from the surface of the earth—, it becomes necessary to discuss and decide "who" within government will work for this not to happen. It is not a matter of finding a super hero. It is a matter of building institutions that survive the pass of time and administrations, and deal with AI from a public policy perspective and not from the lenses of politics or ideology. As history and experience shows, not many governments will get right the "who" behind AI governance and policy. Let's talk about it.
The effective destruction of government effectiveness
Today, many people within government, civil society, think tanks, expert groups and even top executives from big tech firms raise their voice asking governments to regulate or do something about AI's risks and challenges. The latest round included the leaders of OpenAI and Anthropic calling to slow down the pace of AI development—which to a greater extent is in their capacity to do so. The US federal government called for self-regulation and key industry players agreed on voluntary commitments towards AI safety.
Raising your voice and requesting governments to do something is easier and less costly than taking a step further and defining what "doing something" entails. Of course, a reasonable person would say that people in government are the experts and that it is part of their job to tackle issues like those coming from the risks and major threats AI might pose. But, lately those in government stopped being the experts.
In some places like my country of origin and the country where I live, new administrations promised government efficiency and austerity. But when fulfilling their promise, they applied their razor too deep. Yes, they got rid of bureaucratic fat, but they also cut key muscles and gray matter needed to perform vital government functions.
Instead of rebuilding institutional strength, they double down their bet and decided it was enough to govern over the mic or social media. But there is so much an approach like this can deliver. Speeches and tweets cannot substitute critical thinking and hard policy and political work. They cannot design and build effective institutions that deal with serious public policy problems—including fast pacing and changing challenges from AI's development and use.
Why is someone needed to lead AI governance and regulation?
The emergence of artificial intelligence—with its public benefits and risks—, comes at these times of weak institutional capacity and leadership in many governments. This is a serious problem, not only because of the challenges of regulating AI, but also because fundamentals and basic knowledge have been lost in some governments. One of the lost 101 lessons is the understanding that you need institutions to deal with public policy problems beyond the news cycle.
In my book, "Better Regulations, Better AI Results", I dedicate a chapter to discuss the different "whos" responsible for leading efforts across government, implementing solutions and giving political support to initiatives that improve the governance and regulation of artificial intelligence—which are not easy tasks on themselves. At the beginning of that chapter, I included a quote from Loayza and Woolcock, authors of "Designing Good Policies Is One Thing, Implementing Them Is Another." They say:
"Forging and adopting technically sound policies is necessary for successful development, but it is not enough: any policy is as good as its implementation... But without a supporting institutional framework and capable public sector organizations to implement them, even technically sound policies and program are likely to fail."
There are two angles to what Loayza and Woolcock suggest. One is that you cannot get good public policy results without adequate and dedicated institutions responsible for achieving them. Thus, it would be irrational to expect good results in public health, the economy or national security without having in place at least one or more responsible agencies or government bodies for each of these policies.
Applying this idea into the field of AI, it is unreasonable to expect good AI governance and policy results, when a government does not have at least a dedicate agency focused on this disruptive technology with the potential of huge economic, political and social impact—both positive and negative.
The second angle from Loayza and Woolcock is that governments often need more than one institution to achieve policy goals. For instance, attracting investment to a country touches on so many aspects in the public sector that require coordination and collaboration from multiple government agencies—national and subnational—, which need to collaborate to capture investment opportunities. That's part of the "supportive institutional framework" in their statement.
Because AI affects many aspects of people's lives and business activities, a dedicated body for AI governance and regulation might not be enough—making coordination, support and collaboration across government agencies and levels extremely important. This is not a task that happens spontaneously and without great effort and energy.
Wanted! Not a fugitive person, but institutions with AI governance responsibilities
Whether countries need a new AI agency, department, ministry or taskforce depends on their current institutional arrangement. In some cases, an existing government agency or team may have the capacity and bandwidth to lead and coordinate AI governance across agencies and levels of government.
When a new institution is needed, its creation should not be approached as a box-ticking exercise, leading to one-more bureaucratic elephant that moves and goes nowhere.
Even a small team at the center of government can grow and spinoff into an independent public body in charge of AI governance and policies. What will really matters is not the size, name or public announcements of new AI institutions. What really matters are the functions and responsibilities, having a solid foundation that includes a formal mandate, adequate financial and human resources and of course, political support from the President or Prime Minister.
In my book, I talk about these 3 "who's" that need to be in place for a more solid AI governance—regardless of the country, political system or administration:
1. The 'who' leading AI governance.
This is the agency, department or team leading the way, support AI policy implementation and coordinating with other agencies, Congress, state governments and relevant stakeholders. If you assess the challenges regulating AI represents and the functions describe in the previous lines, you may realize this is not a task for an AI Czar or group of 5 to 10 people. The questions for government is whether this role fits in their current institutional structure or whether the challenge call of a new agency.
2. The 'who' implementing the 'how' and 'what' of AI regulation.
At some point, things need to tickle down to be implemented across all government agencies. Because AI is being used for many purposes and activities including health, job selection, loan applications, manufacturing, self-driving cars, psychological advise—and more—, AI will fall under the scope of many regulatory agencies. When AI regulation is needed, these agencies will need to understand the technology and apply good regulatory practice that do both, protect what needs to be protected while not killing innovation with unnecessary requirements. All this will not happen with support from "1" above and adequate human and financial resources.
3. The 'who' letting all of this happen.
In many governments there are pretty good elements on 1 and 2 above. However, they often fail to deliver results because the center of government reacts with friction and reluctant to a policy agenda that requires loosing the leash of control, creating more transparency, dialogue, delegation, cooperation and having less spotlight. Then, the last key ingredient comes from the President, Prime Minister and their closest team, to let a better AI governance happen. It means being a leader who demands less and provides more, so others can do their job and shine.
What some governments are proposing or already doing
Creating institutions resembles more an art than a science. There is no one-size-fits-all approach, as every country has unique way of structuring their national and subnational government for very local-specific reasons. Therefore, the intention of the list below is to show a little of what is happening in some countries, rather than suggestion that these arrangements are better or should be replicated.
Internationally. This remains at the proposal stage. On the sidelines of the most recent United National General Assembly, 22 countries made a joint declaration calling—among other things—to explore the creation of an "international institution, able to set standards, enable verification, and convene states when capability thresholds are crossed." Signing countries were: Norway, Finland, Australia, Bahrain, Canada, Denmark, Estonia, Germany, Iceland, Ireland, Kazakhstan, Kenya, Latvia, Moldova, Netherlands, Singapore, Spain, South Africa, Türkiye, the Untied Arab Emirates as well as the European Commission.
Spain. The Agencia Española de Supervisión de la Inteligencia Artificial (AESIA) is an autonomous agency under the Secretariat of State for Digitalization and AI (SEDIA), within Spain's Ministry of Digital Transformation. AESIA, with more than 50 staff and created by a Royal Decree, is responsible for the ethical and safe use of AI. It has a mix of coordination, inspection, verification and sanctioning powers, and is the oversight backbone of a broader institutional arrangement around AI that includes the Directorate General of AI as well as the International Advisory Council on Artificial Intelligence. Overall, Spain is way ahead other countries and was the first EU member to establish an AI regulatory sandbox.
United Arab Emirates. The UAE institutional arrangement aligns with the National Artificial Intelligence Strategy 2031. Back in 2017, the UAE was the first country in the world to create Ministry of Artificial Intelligence. Recent reforms created a new Artificial Intelligence and Data Authority that reports directly to the Cabinet. The new agency merged 3 bodies: the Office of Artificial Intelligence, Digital Economy and Remote Work Applications; the Digital Government Sector at the Telecommunications and Digital Government Regulatory Authority (TDRA); and the UAE Data Office. The new AI and Data Authority will be responsible for the implementation of the National AI Strategy, will propose laws, regulations, standards and policies and will also lead the government's efforts to transition towards Agentic AI.
Singapore. The country relies mainly on its existing institutional framework and voluntary requirements. The National AI Council, led by the Prime Minister, oversees the country's AI strategy and policy. Also, the Ministry of Digital Development and Information integrates policy efforts, while its Infocomm Media Development Authority (IMDA) leads the nation's AI policy and hosts the AI Safety Institute. IMDA has several voluntary frameworks and guidelines on issues like: transparency for GenAI chatbots, governance frameworks for Agentic AI, legal responsibilities for AI agents and AI safety.
Canada. After having and business-oriented policy, with its new "AI for All" national strategy, Canada is aiming more for specific AI laws and regulations, rather than creating a comprehensive AI Act. Institutionally, in 2025 Canada appointed a Minister of Artificial Intelligence and Digital Innovation, who sits within ISED (Innovation, Science and Economic Development). The country also has a Canadian AI Safety Institute. The overall national coordination is led by the Treasury Board of Canada Secretariat (TBS), which has launched, established and published initiatives like the Directive on Automated Decision-Making, the AI Strategy for the Federal Public Service 2025-2027, the Government of Canada AI Register, the Guide on the Use of GenAI, and the Algorithmic Impact Assessment.
United States. Despite its global leadership on AI, the US lacks institutions that directly deal with AI governance and regulation. Recently, the federal government created the "Super Intelligent Force" and appointed an AI Czar—following government and industry discussion regarding AI safety and the risk of AI going rouge. Without a legal underpinning, the AI Force has 120 days to prepare a report and develop a plan to respond to AI threats, while preventing overregulation or regulatory capture. Nonetheless, debates outside the federal government consider the need to have a dedicated agency.
It's goes beyond the institutions, AI governance needs better leaders
Wanted! This call is not limited to AI institutions responsible for AI governance and regulation. Despite their long-lasting nature, institutions cannot succeed without leadership, "Wanted!" is also a call for for responsible leaders who take a stand for an "AI for all" policy—as Canada did.
An "AI for all" implies that there is space for AI's growth as well as for all the benefits AI firms and advocates have promised. It also means there is space for creating safeguards and protection of things that matter, like people, the environment, jobs, safety and many more.
An "AI for all" means rethinking a global AI race viewed only as a zero-sum game with a winner-takes-all kind of finale, to instead realize AI's greatest potential for good is something to share and nurture. This means looking and engaging with other countries, which may be doing more innovative and interesting things from a governance perspective than the nations at the top of the AI race.
However, many will argue that this is not how life and government works. Maybe this is true. Or perhaps this is the view that a few benefiting from this idea have made sure it prevails in the public consciousness—crowding out the possibilities for transformation and different styles of leadership.
That's why AI governance, policy and regulation demand better leaders who can operate beyond the chains of ideology, politics, economic interests, election cycles and fear. Of course, not every "leader" can act like this. That is why I say that successful AI governance is not something anyone can do. This makes the call "Wanted!" more urgent than ever. Not to find a superhero, but to find someone responsible who can help and lead a collaborative effort to build the AI governance institutions of the future.