The European Commission presented an action plan on Monday to bolster the bloc's cybersecurity defences against risks posed by advanced artificial intelligence models, including systems such as Anthropic's Mythos.
Vice President Henna Virkkunen, speaking in Strasbourg, said ENISA and the Joint Research Centre will build a secure platform to test AI applications in cybersecurity for critical sectors, with a European evaluation capacity to be operational by 2027. The Commission will also work with ENISA to define access guidelines so public and private organisations can use advanced AI models. An EU AI Office will gain oversight powers within weeks to test models directly.
The plan proposes no new legislation or funding, instead accelerating enforcement of existing rules. Virkkunen stressed the urgency of implementing the NIS2 directive, noting some member states are significantly behind. Separately, the European Central Bank has warned banks to submit cyber-defence plans by October.