Countries signing Council of Europe’s AI framework convention should not be tempted by self-regulation option
Reporters Without Borders (RSF) hails the Council of Europe’s Framework Convention on Artificial Intelligence as a significant step forward but warns that it makes the mistake of leaving the door wide open to private sector self-regulation. To combat disinformation effectively, governments must regulate all AI stakeholders, both private and public sector, RSF says.
Adopted last month by the Council of Europe, an international organisation comprising 46 European countries, the binding Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law is an excellent document in many respects, but it does not give itself the means to achieve its goals.
The Convention recognises that AI can be used for disinformation purposes and highlights the risks that AI systems pose to the integrity of electoral processes. But it contains a structural flaw that could render it completely ineffective.
Its mistake is to make a clear distinction between AI systems used by the public sector and those deployed by the private sector for private sector use. In the public sector, members states are required to adopt strict measures to regulate the use of AI, identifying specific risks, including risks linked to disinformation.
But the Convention gives member states a free hand to manage these risks in the private sector as they see fit, thereby making it possible for them to opt for self-regulation, that is to say, to allow the AI industrial sector to regulate itself, using charters or codes of conduct for example. However, AI’s pioneers are private sector companies with private sector clients. Leaving the door open to self-regulation will not address the challenges posed by this new industry.
The self-regulation permitted by this framework convention will not ensure that AI serves the general interest and protects the right to information. Designers of digital services, such as social media, have so far shown no inclination to sacrifice their business interests in order to ensure that their products serve the general interest and the right to information. We call on countries that sign the convention to opt for the imposition of legal requirements rather than taking the road of self-regulation, and we urge them to pass robust national legislation that strictly regulates the development and use of AI, in order to ensure that it is used in an ethical and responsible manner, and to thereby protect the integrity of information and democratic processes.”
RSF’s recommendations for AI that respects the right to reliable information offer a sound framework that overcomes the shortcomings of both the Council of Europe’s Framework Convention and the European Union’s AI Act.
