There's so much noise around this topic that people struggle to understand what it actually requires.
As the European Union's landmark AI Act takes effect, Portugal has chosen not to wait for clarity to arrive from above — it is building that clarity itself. A coalition of researchers, biomedical scientists, and language technologists has launched Bridge AI, an initiative designed to translate sweeping regulatory language into knowledge that citizens, businesses, and policymakers can genuinely act upon. The effort reflects a deeper truth about governance in the age of algorithms: laws alone do not protect people — understanding does.
- The EU's AI Act is transformative legislation, but its complexity risks becoming noise — most people, including those who must enforce it, struggle to grasp what it actually demands.
- Bridge AI unites an unlikely coalition — a computer science institute, a biomedical foundation, and an AI company — signaling that no single sector can decode this challenge alone.
- Five working groups are dissecting real-world AI case studies, exposing both the genuine promise of AI in healthcare and welfare and the ethical and legal fault lines running beneath them.
- International experts and national organizations are converging around a shared deadline: evidence-based recommendations, ready by October 2024, to guide Portugal's implementation of the Act.
- If the project succeeds, Portugal could emerge not just as a compliant member state, but as a European model for turning regulatory ambition into grounded, actionable governance.
Portugal is attempting something most countries have yet to manage: making the European Union's new AI regulation legible to the society that must live under it. The initiative, called Bridge AI, launched in 2024 with an unusual coalition — INESC-ID, a computer science research center; the Champalimaud Foundation, a biomedical research funder; and Unbabel, a Portuguese AI company. Together with over a dozen national partners and international experts including Virginia Dignum and John Krakauer, they are working to answer some of the most pressing questions surrounding the AI Act: How do you build AI literacy across a population? What ethical frameworks should guide implementation? And how do you ensure a law this complex is enforced as intended?
Helena Moniz, the project's coordinator at INESC-ID, frames the challenge directly. AI already shapes decisions in healthcare, finance, employment, and criminal justice — yet the surrounding noise makes it difficult for anyone to understand what the AI Act actually requires. Bridge AI exists to cut through that confusion, offering grounded, evidence-based guidance to citizens, business leaders, and government officials alike.
The project operates in two phases: five working groups analyzing real AI case studies from the Center for Responsible AI, followed by a public-facing translation of their findings. The cases reveal both the genuine potential of AI systems and the ethical and legal complications they introduce — particularly in sensitive domains like healthcare. The goal is not to obstruct innovation, but to ensure that when Portugal implements the AI Act, it does so with full awareness of what is being protected and what is being enabled.
Funded by Portugal's Foundation for Science and Technology and the office of the Prime Minister, Bridge AI is expected to publish its recommendations by October 2024. If it delivers, Portugal may offer Europe a rare thing: a working model for turning complex, continent-wide regulation into something that actually shapes how AI is built and deployed on the ground.
Portugal is trying to do something most countries haven't figured out yet: translate the European Union's sweeping new AI regulation into something a society can actually understand and use. The effort is called Bridge AI, and it launched this year with an unusual coalition behind it—a research institute, a foundation, and a language-technology company working together to close the gap between what lawmakers wrote and what people need to know to live under those rules.
The three organizations leading the project are INESC-ID, a computer science research center; the Champalimaud Foundation, which funds biomedical research; and Unbabel, a Portuguese AI company. They've brought in more than a dozen other national organizations and a roster of international AI experts—Virginia Dignum, John Krakauer, Allaine Cerwonka among them—to tackle a set of urgent questions. How should Portugal build AI literacy across its population and among the people making policy decisions? What ethical frameworks should guide how the country applies the AI Act? What can Europe learn from how other parts of the world are handling AI regulation? And how do you actually make sure a law this complex gets enforced the way it was meant to?
Helena Moniz, a researcher at INESC-ID and the project's coordinator, frames the problem plainly. AI already shapes daily life across sectors—healthcare, finance, employment, criminal justice. The EU's AI Act is genuinely transformative legislation, she says, but there's so much noise around the topic that people struggle to understand what it actually requires. Bridge AI exists to cut through that confusion, to give Portuguese society, business leaders, and government officials the kind of grounded, evidence-based information they need to make real decisions.
The project is structured in two parts. The first borrows from how the European Union itself makes decisions: five working groups, each focused on a specific dimension of the AI Act, each with concrete objectives and deliverables. The second part is about translation—taking what those groups learn and making it legible to the broader public. The working groups are analyzing real case studies pulled from the Center for Responsible AI, scenarios that show both the promise and the peril of AI systems in the world.
Moniz emphasizes that many of these cases demonstrate genuine potential—AI improving health outcomes, enhancing social welfare. But they also surface ethical and legal complications, especially in healthcare and other sensitive domains. That tension is exactly why Bridge AI came into being. The goal isn't to block AI or to make regulation so rigid it strangles innovation. It's to ensure that when Portugal and other countries implement the AI Act, they do it with their eyes open, understanding both what they're protecting and what they're enabling.
The project is funded by Portugal's Foundation for Science and Technology and by the office of the Prime Minister. By October 2024, Bridge AI is expected to publish its findings and recommendations—a roadmap for how a country can build the knowledge and institutions it needs to regulate AI responsibly. If it works, Portugal could become a model for how to translate a complex, continent-wide regulation into something that actually shapes how people build and deploy AI systems on the ground.
Notable Quotes
AI already has major impact on all our lives and across many sectors. It's important to ensure the best understanding and application of the AI Act, which is truly revolutionary legislation.— Helena Moniz, INESC-ID researcher and Bridge AI coordinator
The Hearth Conversation Another angle on the story
Why does Portugal need its own project on this? Isn't the AI Act already written and binding across the EU?
The law exists, yes, but understanding it and actually implementing it are two different things. The AI Act is genuinely complex—it touches everything from healthcare to hiring to criminal justice. Most people, including many policymakers, don't have the literacy to know what it requires of them or how to apply it ethically. Bridge AI is about closing that gap.
So this is really about education and translation?
Partly, but it's deeper than that. They're also analyzing real cases—actual AI systems in the world—to understand what the ethical and legal questions actually look like when you're not in a seminar room. That's where the learning happens.
Who benefits most from this work?
Ideally, everyone. Companies need to know how to build AI systems that comply with the law. Policymakers need to understand what enforcement looks like. And citizens need to know what rights and protections the law gives them. Right now, that information is scattered or missing.
What happens if Bridge AI succeeds?
Portugal becomes a model for how to do this well. Other countries watch, learn, adapt. And the EU gets better data about whether the AI Act is actually working the way it was designed to.
And if it doesn't?
Then you have a law on the books that nobody really understands, and enforcement becomes arbitrary or ineffective. That's the risk they're trying to prevent.