Technology: The Council of Europe adopts the Artificial Intelligence – Digital Agreement

Countries all over the world can join

Once signed, not only Council of Europe countries will be able to join the agreement, but also countries around the world. Therefore, Foreign Minister Annalena Baerbock described the agreement as a “milestone.” The agreement “also shows that the Council of Europe has global appeal because other countries from very different continents are joining this first agreement on artificial intelligence,” the green politician said.

The 46 member states of the Council of Europe include all 27 EU countries, as well as countries such as Great Britain and Turkey. It is therefore responsible for 680 million people – from Greenland to Azerbaijan. The United States, Israel and Canada also participated in the negotiations as observer states. Therefore, there was great hope that the agreement would have a wide-ranging impact, and that it would include, for example, powerful technology companies in the United States of America. Whoever signs the agreement is bound by it.

Agreement with several exceptions

According to the Council of Europe, the contract sets out, among other things, requirements for transparency and monitoring, for example when content is generated by artificial intelligence. Countries must also ensure that AI systems respect the prohibition of discrimination and the right to privacy, he said. It must also ensure that AI systems are not used to undermine democratic processes. In order to achieve these goals, States should establish independent supervisory mechanisms.

However, the requirements do not apply to matters of national security and defense. When regulating the private sector, states can also choose whether to adopt their own measures rather than the Convention's regulations. The Council of Europe said this was necessary because of different legal systems. However, critics complain that this weakens the agreement and gives countries and companies too much freedom.

Angela Müller of Algorithm Watch called it a “valuable, charismatic signal” that the Council of Europe has recognized that rules on dealing with AI are essential to protecting human rights. But the contract leaves a “bitter aftertaste” because it does not do justice to these goals. “Although there is no shortage of evidence about how technology companies influence the formation of public opinion through social media algorithms or deepfake generators, the Council of Europe leaves it up to countries whether they want to adopt soft measures or binding laws,” Müller told the German news agency DPA. (DPA): “It trusts that pure self-regulation by companies will be sufficient to protect human rights and democracy.”

Data Protection Officer: Red lines are missing

The European Data Protection Supervisor had already warned in the final stage of negotiations in March that the agreement might become a “missed opportunity.” The main criticism was that the draft lacked red lines for some AI applications. There are concerns that the agreement is too broad and will therefore be applied differently.

In the coming years, Germany must now implement the EU AI law and the Council of Europe AI Convention and integrate them into national law. “The federal government can, at least in part, overcome the failures of the Council of Europe and the EU by banning some AI applications, such as facial recognition in public spaces,” says Algorithm Watch's Müller.