The European Union has agreed on unprecedented legislation at the global level to regulate artificial intelligence (AI Act) after three days of intense negotiations between Member States and the European Parliament.
An agreement to regulate artificial intelligence in Europe
The EU’s co-legislators have reached a political agreement on a text that should promote innovation in Europe while limiting the possible excesses of these highly advanced technologies.
“Historic! The EU becomes the first continent to set clear rules for the use of AI,” said European Commissioner Thierry Breton, who initiated the project presented in April 2021. Since then, discussions have dragged on. The latest round of negotiations, which began on Wednesday afternoon, lasted nearly 35 hours.
The process was impacted at the end of last year by the appearance of ChatGPT. This system, like those capable of creating sounds or images, has revealed to the general public the immense potential of generative AI. But it has also exposed some risks. For example, the dissemination of fake photographs, which are larger than life, on social networks has warned of the danger of manipulating public opinion.
This phenomenon of generative AI has been integrated into the ongoing negotiations at the request of MEPs who insist on a specific framework for this type of high-impact technologies. In particular, they called for more transparency about the algorithms and giant databases at the heart of these systems. Member states feared that over-regulation would nip in the bud their fledgling champions, such as Germany’s Aleph Alpha and France’s Mistal AI, by making development costs prohibitive.
The political agreement reached must be complemented by technical work to finalize the text. We will carefully analyze the compromise reached today and ensure in the coming weeks that the text preserves Europe’s ability to develop its own artificial intelligence technologies and preserves its strategic autonomy,” said French Digital Minister Jean-Noël Barrot.
A trade-off for generative AI
On generative AI, the compromise provides for a two-tier approach. Rules will be imposed on everyone to ensure the quality of the data used in the development of algorithms and to verify that they do not violate copyright law. Developers will also need to ensure that the sounds, images, and texts produced are identified as artificial. Enhanced constraints will apply only to the most powerful systems.
The core of the project consists of a list of rules imposed only on systems deemed “high risk,” mainly those used in sensitive areas such as critical infrastructure, education, human resources, policing, etc. These systems will be subject to a series of obligations such as human control over the machine, the establishment of technical documentation, and the implementation of a risk management system.
The legislation provides a specific framework for AI systems that interact with humans. It will oblige them to inform the user that he or she is in contact with a machine.
Fines of up to 7% of turnover
Bans will be few and far between. They will concern applications that are contrary to European values, such as citizen rating or mass surveillance systems used in China, or the remote biometric identification of people in public places to avoid mass surveillance of populations. On this last point, however, States have obtained exemptions for certain law enforcement missions, such as the fight against terrorism.
Unlike the voluntary codes of conduct of some countries, European legislation will be equipped with means of monitoring and sanctions with the creation of a European AI Office within the European Commission. It will be able to impose fines of up to 7% of turnover, with a minimum of €35 million, for the most serious infringements.