Despite pressure from some EU countries, the European Commission is still trying to prevent private companies from being excluded by default from the first international treaty on artificial intelligence.
The Council of Europe (CoE), an international human rights body with 46 member states, is close to finalizing a treaty on artificial intelligence, human rights, democracy and the rule of law.
The United States, home to some of the world's largest AI companies, has called from the beginning to exclude the private sector from the treaty, which would be binding on signatories if ratified.
Although the United States is not a member of the CoE, it participates in this process as an observer. In other words, Washington does not have the right to vote, but it can influence the debate by saying it will not sign the treaty.
Pressure on the United States and other countries to support a more narrowly scoped treaty, despite contrasting with the CoE's internal guidelines on stakeholder participation, meant that the entire drafting procedure was kept confidential and that NGOs were excluded from the process. It was enough to be excluded.
broad scope or broad participation;
By contrast, the European Commission, which represents the EU in negotiations, opposes this carve-out to the private sector. Two weeks ago, Euractic revealed an internal memo that said “unions should not agree to alternatives that limit the scope of the treaty.”
However, at a subsequent meeting of the Telecommunications and Information Society Working Group, the technical arm of the EU Council of Ministers responsible for digital policy, several member states signaled to the European Commission a more flexible stance regarding the scope of the treaty. I asked.
Particularly for countries such as Germany, France, Spain, the Czech Republic, Estonia, Ireland, Hungary, and Romania, the purpose of this treaty is to reach a global agreement, and therefore a broader treaty with more signatories The priority should be to secure as many signatories as possible. International support is limited.
The position of the bloc, which is made up of 27 of the 46 member states of the Council of Europe, could itself upset the balance within human rights institutions, where decisions are taken by consensus. .
The EU's global ambitions
Restricting the scope of the Treaty would be a serious blow to the European Commission's global ambitions. The Commission sees the treaty as a way to make the EU's AI law, the world's first comprehensive law on artificial intelligence, a global benchmark in this field.
In fact, the European Commission's mandate to negotiate on behalf of the EU is based on AI law, and the EU's executive branch, despite the fact that the two approaches are different, has no direct dispute in areas where there is no direct conflict. has shown little desire to go beyond AI regulation. Noticeably in nature.
As part of its alignment with the AI Act, the European Commission is seeking broad exemptions for the use of AI in national security, defense and law enforcement. Therefore, if the treaty were limited to public bodies only, and there were these carve-outs, very little would remain.
Additionally, Euractiv understands that such a significant watering down of the AI Treaty, after several years of efforts by the countries involved, may hinder future efforts in this area.
Nevertheless, the final word on this issue is yet to be said. The committee's original memo was shared in preparation for the plenary session of the CoE's Artificial Intelligence Committee, which began on Tuesday (23 January) and runs until the end of this week.
Participating countries are expected to reach an agreement on the scope at this General Assembly, with a view to formally adopting the treaty at the ministerial level in May. Talks have so far been inconclusive, including an informal one last week, with a final decision expected only on Friday.
Opt-out option
Ahead of the plenary session, the European Commission shared an updated Memorandum of Understanding with EU national delegates stating: “The EU should continue to pursue its goal of ensuring comprehensive coverage of the Convention, covering both public and private actors.'' ' he claimed.
Notably, “in order to maintain the international character of the Treaty, the EU may be prepared to consider the possibility for States Parties to make reservations and exempt private actors from the obligation to apply the Treaty.'' A paragraph has been added emphasizing “No.” Subject to certain conditions and restrictions, we do not act on behalf of or procure AI systems for public authorities. ”
The European Commission's proposal appears aimed at addressing the US government's argument that it cannot commit to anything beyond its domestic legal framework.
In October, US President Joe Biden signed an executive order establishing a framework for federal agencies to purchase and use AI tools safely and responsibly, referring to companies that do not work with the public sector. There is.
More precisely, the European Commission is proposing a temporary limited “opt-out” option, which can be amended at any time and has certain guarantees that it will not be abused. This approach would be the opposite of what the US government has proposed, which would give signatories the possibility to “opt in” and exempt the private sector by default.
Still, the original “opt-in” option was designed to avoid the embarrassment of the U.S. government having to exempt private companies from human rights treaties. Euractic understands that if an “opt-out” approach is included in the final document, Israel and Japan will not sign, while the UK and Canada will abide by the US decision.
[Edited by Nathalie Weatherald]
