AI non-proliferation: Parliament takes the lead, says Gozi

The vote in the Committee on Constitutional Affairs (AFCO) on 1 October 2026 puts the principle of non-proliferation of high-risk and harmful artificial intelligence systems into the Parliament's report on the institutional aspects of AI. The amendment was tabled by Sandro Gozi, Renew Europe MEP and Secretary General of the European Democratic Party.

"The European Parliament's Committee on Constitutional Affairs has adopted my amendment on the non-proliferation of high-risk and harmful artificial intelligence systems," Gozi said. "The principle is now part of the report on the institutional aspects of AI and must underpin any international convention or treaty on the matter. Parliament is thus putting itself forward to lead the effort to fill a gap in global rules that currently leaves systemic risks unchecked."

Learning from the nuclear age

Gozi compared the issue to the nuclear era. "With nuclear power, the world learned late, and at great cost, that some technologies must be governed before they get out of hand. With AI, we can get there in time."

He described a global contest between three models: the American, the Chinese and the European. In his view, the European model is "the only one that holds innovation and rights together". Taking it to the United Nations, he argued, "means defending our democracy against interference and strengthening our competitiveness."

Europe should hold the pen

The rules on AI will be written one way or another. The question is who writes them. "Artificial intelligence offers immense opportunities and equally great risks," Gozi concluded. "The world's rules on this technology will be written either way: it is better that Europe holds the pen."

For the EDP, this is the right approach for a Union that wants to be a global player: protect citizens and democratic institutions without holding back innovation, and make the European way of governing technology a standard that others want to adopt.