In the thick of things. Since 14 June, applications from Luxembourg have been open with a view to joining the Sandkëscht programme run by the National Commission for Data Protection (CNPD). “We want to include a wide selection of participants from both the private and public sectors,” stated the CNPD, “as well as organisations of all sizes.” For the purposes of this 100% AI sandbox, the CNPD said that it is “particularly interested [in] issues where there is uncertainty around the interpretation and implementation of regulations, and in particular the GDPR”.
Not to be outdone, the Luxembourg Institute of Science and Technology has designed its own sandbox in order to “help the Luxembourg ecosystem raise its profile and increase its capacity to improve the reliability of AI models and reduce the risks for their users”. The research institute described this sandbox as “a tool to help stakeholders prepare when the law on AI comes into force, and the details of the regulatory sandbox for AI become available”.
An open exchange with citizens and businesses in regulating a subject as dynamic as AI will be crucial.
In the wake of the European Council’s adoption--on 21 May--of the AI Act, the world’s first regulation governing the use of artificial intelligence, the players in the market are getting their act together. Notwithstanding the warning issued by the European Court of Auditors, which in a recent report pointed out the limited scope of the investments made to date and the delays in comparison with China and the US, the EU is loud and clear in its claim to pioneering status, and is urging the 27 member states to get on with it without delay. This urgent need to speed up is reflected in the words of the deputy minister for media and communications, (CSV). She is determined to transpose the European texts into grand-ducal law as quickly as possible, in order to “ensure the rapid adoption of AI in the real Luxembourg economy”.
Opportunities
“AI and the AI Act already occupy a strategic place in the near future of our country’s competitiveness. We cannot miss the opportunities presented by the rapid adoption of AI in our economy, for example in financial services. The government must therefore identify the current barriers and rapidly implement the AI Act among the other frameworks of the digital rulebook, so as not to hold back cutting-edge projects. It’s also a question of alignment with our ambitious Digital Decade trajectories, which I’m coordinating with the other ministries involved,” said Margue, a lawyer by training. “Yes, the digital rulebook is a complex package, I’ve observed that as a lawyer.... At the same time, our intention is to reduce the corresponding red tape as much as possible and not to forget the reasons why we set up this framework: a single digital market that works efficiently.”
Margue continued: “We see opportunities above all in the AI Act, such as the establishment of a regulatory sandbox. Also, the risk-based approach, which ensures that regulation is targeted at really crucial scenarios, for example in human resources, is a positive approach, while innovation is actively supported by the AI Act. At the same time, an open exchange with citizens and with businesses in the regulation of a subject as dynamic as AI will be crucial for the regulators involved.”
‘Political priority’
“A holistic AI strategy is, for me, a political priority", Margue stressed. As such, the officials are “currently revisiting the government’s vision for AI, which dates from 2019. As strategy coordinator, I launched this work in the spring with my colleagues responsible for research, digitalisation and the economy, and ”, Margue said. What are the urgent issues? “We will certainly be focusing on the economic sectors that are strategic for the country, and the same goes for research priorities.” sFintech, space and health are among the sectors that will be particularly concerned.
Whatever the case, “what counts for me,” concluded Margue, “is a regulator open to dialogue with innovative companies, an overview of the economic transition triggered by AI and a particular expectation linked to citizens’ rights. Citizens, like businesses, will always need to be guaranteed that they can interact with a human being and not with an algorithm. The principle of human-centric AI will be respected in Luxembourg”. The roadmap is taking shape.
Originally published in French by and translated for Delano