The growing technological gap between Europe and the United States is beginning to take on a dangerous security edge for public administrations and for companies across the Old Continent, as the risks deepen. Beside the distrust of Donald Trump’s unpredictability and the worry that the US president might weaponize Europe’s digital dependency for political or commercial purposes, there is now the threat posed by the trial launch of Claude Mythos Preview. This new artificial intelligence model from the American company Anthropic is not, according to experts, another leap forward; rather, it opens a completely new landscape in which the risks of cyberattacks against any firm or government body—lacking the resources or know-how to defend themselves—could multiply exponentially.
Alarm bells have sounded in Brussels and the European Commission is negotiating with Anthropic, so far unsuccessfully, to allow European businesses or agencies to access the model before its final market release. In theory, the European AI Regulation enables Brussels to verify the safety of new AI systems to determine whether they present “unacceptable risks” or to impose conditions on those deemed to pose a “systemic risk”.
“The European Commission’s wavering to curry favor with Trump and not upset his capricious Administration also makes Brussels prudence itself more pronounced”
But that rule, unveiled with much fanfare in 2024 as a global first, is far from having been tested in reality; it has already needed one amendment and the implementation timelines have been pushed back beyond what was initially planned. The Commission’s posturing in order to appease Trump and avoid angering his unpredictable administration also makes Brussels exercise extreme caution in the increasingly tense relations with the American multinationals that lead the advances in the digital space.
The European Parliament plenary will address the risks posed by AI models like Mythos for software programs and for the security of Europe’s critical infrastructures, as well as the possible consequences for the European economy at large. Eurodeputies, who do not hide their concern, will demand explanations from the European Commission about the access of European authorities to AI models in testing and about how the current legislation is being applied. There is a sense that the speed of AI progress in the United States is leaving Europe behind, a continent eager to reap the benefits of the new technology but struggling to cope with its risks.
“The ECB warns that the new AI model will readily enable cyberattacks that until now were within the reach of only highly specialized individuals”
At the forefront of the new digital frontier stand European banks, long accustomed to facing threats, scams, and digital intrusions that will seem child’s play compared with what may arrive thanks to Anthropic’s AI. “This is an urgent situation; we are not facing a distant scenario,” warns Frank Elderson, vice-chair of the ECB’s Supervisory Board and a member of the ECB Executive Board, in a kind of interview published by the bank itself last Wednesday. Elderson urges euro-area financial institutions to shift to a preventive culture to tackle even the vulnerabilities that had previously appeared harmless or posed no imminent risk.
The ECB warns that the new AI model will enable cyberattacks to be organized with ease that until now were the preserve of highly specialized individuals or required days of work by a team of IT experts. In Europe, the risk is even greater because Anthropic refuses to share its model with companies or authorities on this side of the Atlantic, which will limit their ability to establish defense mechanisms or anticipate potential attack paths.
But the ECB also cautions that not only banks are at risk, but also “other core infrastructures”, ranging from the power grid to telecommunications. And beyond the financial sector, all kinds of services and utilities—ranging from healthcare to water companies or distributors—could be exposed to a new generation of cyber threats for which Europe does not appear prepared.
“Our economic future has never before in our recent memory depended so much on a single technological transformation”
