Faced with the success of AI, in particular with ChatGPT, the CNIL announced the creation of an artificial intelligence service (SIA) intended to oversee the use of this technology and to respond to the problems it raises.
Artificial intelligence is currently experiencing incredible technological advances. It is enough to see the success of ChatGPT and the upheavals it produces to understand that this revolution will concern absolutely all sectors. But with increasingly sophisticated tools and software, it is normal to worry about possible problems, both ethical and technical, that the generalization of artificial intelligence could cause. Especially since its operation is generally based on machine learning, and therefore on the processing of a large amount of data, including personal data, and that their exploitation entails risks for privacy and other fundamental rights.
Faced with its new problems, the National Commission for Computing and Liberties (CNIL) has decided not to wait for the European Union to finally agree on the amendments to be made to the Artificial Intelligence Act, intended to regulate the use of artificial intelligence in Europe, to address the issues that the subject involves. This is why she announced in a statementon January 23, 2023, the creation of an artificial intelligence service (SIA) which will have to try to answer the various questions and above all to frame the use of AI in the future.
SIA: protecting privacy and regulating the uses of AI
The creation of the SIA follows the proposal of the Council of State in August 2022 to make the CNIL the regulator of artificial intelligence, an issue on which it has been working since 2017. Its goal: “strengthen the CNIL’s expertise on these systems and its understanding of the risks to privacy while preparing for the entry of the European regulation on AI”. It is also a way to look at the transparency and understanding of a technology “very often perceived as a black box” in order to ensure a balanced regulation and to allow the actors of the sector and the CNIL to control the risks for the private life.
Just look at the questions raised by image generators, accused of infringing copyright by training and “copying” artists’ images that are on the Internet and are not free of rights. Speaking of copyright, who owns the created image, the AI or the user? And that’s without counting on the transparency of using such a tool, whether for text, illustration or anything else. How do you know if it was produced by a human or by an artificial intelligence and, if so, what value should be given to it? The new Commission service will have its work cut out for it!
The SIA is currently made up of specialized lawyers and engineers, for a total of five people directly attached to the CNIL’s technology and innovation department, under the leadership of Bertrand Pailhes, a former national coordinator for the artificial intelligence within the DINSIC (Interministerial Directorate for Digital and State IS). The CNIL has already launched work on machine learning databases to provide recommendations and concrete answers to professionals and to give them a framework for analyzing the regulations on the protection of personal data in order to guarantee the security of those of the users.