European Parliament Divided on How to Govern AI At Home and Abroad

European lawmakers want global AI rules but clash over regulation at home.

The European Parliament debated global AI governance in Strasbourg on Thursday, calling for deeper international cooperation to manage AI risks, including through engagement with the United States and China. But the debate also exposed divisions over how far governments should go in regulating AI and how much responsibility should fall on the companies developing it.

The disagreement extends to Europe’s own approach. Brussels defended its decision to withdraw the AI Liability Directive after lawmakers called for its revival.

Addressing the plenary on Thursday, European Commissioner for International Partnerships Jozef Síkela argued that international engagement was essential because "artificial intelligence knows no border." He pointed to existing forums for joint discussions, such as the G7, G20, the United Nations AI summits, OECD, Council of Europe, and the Network of AI Safety and Security Institutes. Nevertheless, the EU wants “to team up and to advance work on model evaluation, verification, peer-reviewing, AI security, and much more."

The debate showed broad political support for international cooperation on AI, including with the United States and China, with several lawmakers comparing the technology's potential risks to those of nuclear weapons. "We have to start talking to people that we don't agree with, and that means talking to China,” said Irish lawmaker Michael McNamara, rapporteur of the AI Omnibus, adding the EU also needs to speak with “sometimes allies in the United States.”

German lawmaker Sergey Lagodinsky backed that approach, proposing an agreement built on the Hiroshima AI Process that would establish binding rules and include both the United States and China. “We need a strategic AI restraint treaty, like we used to create for arms control,” he said.

But lawmakers were divided over how to achieve that cooperation and whether Europe should prioritize tighter safeguards or the competitiveness of its AI industry.

For Greens and Socialists, leaving the rules to major AI companies would be a mistake. German MEP Alexandra Geese, from the Greens, warned that technology "cannot be governed by a blog post," referring to warnings from Anthropic CEO Dario Amodei about the risks of advanced AI, which was echoed by Commission President Ursula von der Leyen's State of the Union speech. Geese also cited former OpenAI and Google DeepMind employees who warned about how financial incentives lead to avoiding oversight.

Sharing some of those concerns, Maltese Socialist, Alex Agius Saliba rejected self-regulation. “With the AI Act and the GDPR, the European Union has led the way on AI governance. Now we must turn that leadership into global action by convening a Global Forum that brings together the US, China, and other major technological powers,” he said. However, he criticized both the withdrawal of the AI Liability Directive and proposals to ease the use of personal data for AI training, discussed in yesterday's Council position on the Data Omnibus.

However, the Swedish center-right lawmaker and co-rapporteur on the AI Omnibus, Arba Kokalari, accused Greens and Socialists of living on a "different planet" for demanding more regulation while European developers fall behind their American and Chinese competitors. "If we want to shape the global rules of AI, we must also lead in developing AI," she said. Lagodinsky pushed back, pointing to reports of autonomous AI agents escaping their testing environments and increasingly sophisticated hacking capabilities as reasons to strengthen safeguards.

The debate over safeguards also produced calls for an international oversight mechanism. German conservative Axel Voss proposed an international inspection regime modeled on the International Atomic Energy Agency, including mandatory testing before AI systems enter the market, incident notification and liability requirements. Earlier in October, the Committee for Constitutional Affairs backed a proposal calling for an AI agreement inspired by nuclear non-proliferation.

The liability question at home

The calls for international cooperation came alongside renewed pressure on the Commission to reconsider its decision to withdraw the AI Liability Directive, which lawmakers argued would have helped people seek compensation when AI systems cause harm.

Síkela said the Commission had supported the directive but faced resistance from national governments." There was a big reluctance by the Council to take this forward, there was no agreement between the co-legislators, and it was not expected that an agreement would be reached soon," he said. He pointed to the AI Act and the revised Product Liability Directive as legislation that already covers certain aspects of liability.

But critics argue those laws do not resolve all the problems victims face when trying to prove that an AI system caused harm.

In a recent interview with Tech Policy Press, Italian Member of the European Parliament Brando Benifei, one of the AI Act's lead negotiators, said the regulation does not provide a general civil liability regime because fines may sanction the infringements without compensating victims. The directive’s withdrawal leaves unresolved difficulties in proving fault and causation, particularly when AI systems make autonomous decisions.

“The problem is not the technology. The problem is big AI companies being reckless by chasing corporate greed,” said Dutch Green MEP Kim van Sparrentak. She argued for holding AI companies to the same standards as other manufacturers. “If a phone charger explodes in your house, we never call it innovative. We call it cheap, dangerous junk, and we hold the manufacturer liable for damages. So why do we let Silicon Valley get away with it?”