On Tuesday 19 May, the main political groups in the European Parliament warned against the risks posed by the European Union’s technological dependence on foreign companies when it comes to advanced artificial intelligence (AI) models such as Anthropic’s Mythos (see EUROPE 13866/26) and GPT-5.5 developed by OpenAI, calling on the EU to invest in European solutions before it is too late.
The EPP, S&D, Renew Europe and The Left groups called for a new, dedicated European strategy to deal with these risks, during a debate in the European Parliament plenary session on cybersecurity in the EU.
“We need a European strategy dedicated to AI-powered cyber threats”, warned Michał Wawrykiewicz (EPP, Polish). And he stressed: “What is at stake is our public safety, our economic stability and the resilience of our democracies”. “We must give priority to our European companies and ensure that we only do business with US technology companies on the basis of reciprocity, not dangerous dependencies”, added Aura Salla (EPP, Finnish).
On the Social Democrat side, Portugal’s Ana Catarina Mendes also warned: “Any strategy we develop must invest in Europe’s cybersecurity resources and in high-level cloud computing, as well as in the protection of critical infrastructures”. “We need to have equivalent firepower”, added her Italian colleague, Brando Benifei, referring to Mythos.
“Europe needs to focus on preparedness and resilience. The money should be spent on infrastructure protection, cybersecurity testing and practical exercises”, said Jaroslava Pokorná Jermanová (PfE, Czech Republic).
On the Conservative side, Beata Szydło (ECR, Polish) regretted that the EU was now “losing out” in the AI race. Her Italian colleague, Alessandro Ciriani, warned that Europe’s competitors are “advancing rapidly” in the field of AI. “We need advanced European models for AI, the cloud, computing power, secure solutions and cybersecurity”, he added.
“We need a European strategy to deal with these particular AI threats and create a European system to reduce our external dependence too”, added Oihane Agirregoitia Martínez (Renew Europe, Spanish). Her Dutch colleague, Bart Groothuis, pointed out that Europe has “very little time to build” its own systems, “a few months at most”. He called on the EU to transform the European agency ENISA into a “task force dedicated to AI cybersecurity” and to “establish a certification system for AI-related cybersecurity tools so that companies know what to trust”.
Speaking on behalf of the Greens/EFA group, Czech MEP Markéta Gregorová deplored Europe’s “strategic dependence” in this area, saying that Europe must “mobilise” its market and regulatory power, and its security intelligence capabilities to obtain access to the Mythos model “at any cost and as quickly as possible” and “harmonise” a single cybersecurity market. “We need European AI defence troops, a European AI defence shield built in Europe and powered by European technologies like Mistral AI, controlled by democracies”, added Sergey Lagodinsky (Greens/EFA, German).
“We really need a dedicated AI cybersecurity strategy to counter these threats”, added Elena Kountoura (The Left, Greek), adding that “models with cyber offensive capabilities must not be freely accessible without supervision”. For her part, Leïla Chaibi (The Left, France) warned that European legislation “is not sufficient” to deal with the risks posed by these advanced AI models. She called for investment in companies like Mistral AI, which is developing a European solution, before it is “too late”.
Her Belgian counterpart, Marc Botenga, criticised the EU for having “directly and indirectly financed” companies such as NSO, Pegasus and Paragon, “which develop spyware for Israel, that is then used against our journalists and infrastructure”.
Finally, Milan Uhrík (ESN, Slovakian) said that the EU should “invest in European companies to get them off the ground”. He criticised the fact that the Commission’s age verification application for social networks “was hacked in two minutes”, warning that if the digital wallet was hacked, “people could lose all their savings, all their money”.
“We will see more and more models of this type coming onto the market from other parts of the world”, noted the European Commission’s Executive Vice-President, Henna Virkkunen, during the debate.
She acknowledged that new AI models are “significantly (lowering) the barriers to entry for sophisticated attacks” and that system vulnerabilities are now being detected “at an unprecedented rate”, surpassing “any human expertise in this area”. However, the Commissioner insisted that current European legislation “already provides us with the legal framework and operational tools needed to deal with these risks”, such as the revision of the Cybersecurity Act, which is awaiting agreement between the co-legislators, as well as the NIS2 Directive and the AI Act.
But this legislation “must now be fully implemented. At the same time, we must continue to strengthen our own European capabilities”, she stressed.
“In order to strengthen our preparedness, in the coming weeks I will be presenting a list of actions bringing together the best expertise in AI and cybersecurity and, in doing so, we will be working with partners who share the same values. If more needs to be done, we are - of course - prepared to consider it”, she concluded.
The Cypriot Deputy Minister for European Affairs, Marilena Raouna, whose country currently holds the Presidency of the EU Council, said that cooperation with industry is “essential” both to “understand technological trends, but also to shape them in line with our values and security”.
She pointed out that from 2 August, under the AI Act, the Commission will be able to demand transparency and risk mitigation measures from these companies regarding their advanced AI models. She also stressed that the new Cybersecurity Act will “considerably” strengthen European cybersecurity. And she believes that Europe must “strengthen” its technological autonomy in AI, referring to the future package on technological sovereignty announced by the Commission (see EUROPE 13870/9). (Original version in French by Ana Pisonero Hernández)