In context: The Council of Europe is an international organization established in 1949 to uphold and protect human rights, democracy, and the rule of law in Europe. Founded in the aftermath of World War II, the CoE has 46 member states and is distinct from the European Union, despite sharing similar goals.
While Geoffrey Hinton, the man credited with creating modern neural network algorithms, is urging governments to prepare for AI advancements with universal basic income measures, the Council of Europe is taking a different approach to address AI threats and opportunities. The European organization has adopted a new treaty, marking the first international agreement designed to manage AI systems and potentially avoid the worst consequences for humanity.
Prepared by the CoE's Committee on Artificial Intelligence (CAI), the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy, and the Rule of Law sets a novel legal framework for countries working with AI algorithms. The treaty aims to address the risks associated with AI and chatbot services while promoting the benefits of "responsible innovation" in AI technology.
The convention includes contributions from the CAI, an intergovernmental body that brought together the 46 CoE member states, the European Union, and 11 non-member states, including the US, Australia, Argentina, and Japan. The treaty covers both the public and private sectors, including companies acting on behalf of the former. Interested parties in the private sector can choose to directly comply with the convention provisions, the CoE said, or take "other measures" that achieve the same outcome while preserving other international obligations.
Specific transparency and oversight requirements have been prepared for different contexts and risks, the organization said, including the proper identification of AI-generated content such as deepfakes or "art" pieces. Risks associated with these contents must be prevented, detected, and mitigated, assessing the need for a moratorium, a ban, or other "appropriate measures" when AI systems threaten the CoE's standards for human rights.
CoE members must adopt specific measures to prevent the misuse of AI systems from undermining democracy, preserving democratic institutions and processes, the principle of separation of powers, and judicial independence. AI-based deepfakes have already become a significant concern for many of the world's major election events scheduled for 2024, including the EU Parliament election and the US presidential election.
The CoE's new convention requires that each party (including member states) establishes an "independent oversight mechanism" to ensure compliance with the new rules, although national security interests and national defense matters will be exempt. The convention itself will be signed on September 5 in Vilnius during a conference of the EU's Ministers of Justice.