When the State Wakes Up Before the Market: AI, Governance, and What We Still Have to Build
- Ricardo Brasil
- Mar 1
- 3 min read
I must admit that February 2026 was a month filled with meaning for me. In just a few days, I experienced two moments that, together, say a great deal about the stage we have reached in the relationship between technology, the State, and society in Brazil.
The first was the publication of RFB Ordinance No. 647 by the Brazilian Federal Revenue Service, which formally establishes a policy for the use of artificial intelligence with a focus on responsibility, transparency, and human supervision. The second was the launch of ENIA, the National Artificial Intelligence Strategy 2026–2029, a project I have been closely involved in from the inside, through the Artificial Intelligence Governance and Management Group (GGIA), coordinated by Professor Dr. João Souza Neto, to whom I am genuinely grateful for the privilege of contributing to this work.
Two initiatives that, individually, would already be significant. Together, they form a signal that deserves far more attention than the technology news cycle usually gives.
What the Federal Revenue Service Did That Seems Obvious — But Isn’t
The Federal Revenue Service ordinance is not a statement of intent. It defines concrete rules: mandatory human supervision in all decisions, prohibition of autonomous AI use to decide processes, requirements for explainable and auditable systems, and strict protection of sensitive data. It seems basic. And that is precisely why it stands out: in an environment where technology companies launch AI products without even publishing their internal usage policies, a Brazilian public authority stepped forward and put the rules on paper before moving ahead.
“There is no responsible use of artificial intelligence where there are no clearly defined limits. Security is not an obstacle to innovation; it is a condition for it.”
This has a name: responsible governance. And that is precisely what ENIA defends as its core principle — human-centeredness, transparency, and accountability. It is not a coincidence. It is coherence.
What ENIA Represents, Beyond the Document
Participating in the construction of ENIA through the GGIA was, for me, an ongoing exercise in productive tension. On one side, the urgency of not falling behind in a global race where every week brings a new model, a new product, a new promise. On the other, the awareness that speed without direction is nothing more than well-accelerated chaos.
The strategy we launched for 2026–2029 does not ignore this tension. It proposes a governance model inspired by the most advanced practices in the world, with risk-based regulation, a decentralized national system bringing together government, sector regulators, and civil society participation, and alignment with international frameworks such as the European AI Act and the U.S. NIST framework. But it does so with its eyes firmly on Brazilian reality: the need for technological sovereignty, the importance of inclusion, and the challenge of capacity building.
The Reflection That Remains
But I would be dishonest if I ended here, simply applauding. There are questions we must continue to ask — including within the governance groups themselves.
The first concerns speed. Documents take time to become practice. And while ENIA sets guidelines through 2029, the market launches and discards technologies in six‑month cycles. How do we ensure that the strategy is not outdated at birth?
The second concerns enforcement. A beautifully written policy without oversight mechanisms is merely a well-crafted intention. The Federal Revenue Service, so far, seems to have understood this by defining not only what, but also how and by whom. ENIA faces the same challenge: turning principles into verifiable practices.
The third, and perhaps the most uncomfortable, concerns who is not in the room.
National strategy documents tend to bring together government, academia, and large corporations. But the AI that will change people’s lives is the one that arrives on the informal worker’s phone, the algorithm that decides credit, the system that screens job candidates. Are these people represented in the decisions we are making?
I do not raise these questions to diminish what has been achieved — quite the opposite. I raise them because I believe it is precisely this critical exercise that distinguishes a national strategy from a celebratory document.
Brazil is, in fact, waking up to AI governance. The Federal Revenue Service has shown that it is possible to act responsibly before being forced to do so. ENIA has shown that it is possible to think systemically and in the long term. The next step — and the most difficult one — is to turn all of this into concrete reality for those who need it most.
This work continues. And it is good that it continues with difficult questions on the table, and with people like Professor Dr. João Souza Neto willing to lead this path with seriousness and purpose.
Until next time.
Check out my other articles:
Comments