After 15 hours of negotiations, Europe has reached a
provisional agreement regulating Artificial Intelligence (AI) through the
adoption of a series of laws. Among other guidelines, the document requires
models such as ChatGPT and AI systems to provide transparency
obligations before being implemented on the market, by means of technical
documentation and legislation relating to copyright. Models that pose a
systemic risk will be subject to testing and evaluation in order to guarantee
cybersecurity and sustainability. The agreement also mentions prohibitions,
namely in the field of cognitive-behavioural manipulation and the collection of
face images from the Internet. Fines for violation range from 7.5 to 35 million
Euros. The European Union is thus a pioneer in setting limits for Artificial
Intelligence.
