The US government halts the launch of Anthropic’s advanced AI models; alarm bells ring in Brussels

In its statement, Anthropic itself admitted that, ‘as we have stated publicly, we believe the Government should have the power to block unsafe applications, as part of a regulatory process that is transparent, fair, clear and based on concrete technical data’. However, it went on to note that “this action does not comply with those principles”.

The announcement therefore marks the latest clash between Anthropic and the US government. The Pentagon classified the company this year as a risk to the supply chain, believing it to pose a threat to US national security: a designation, normally reserved for foreign adversaries, which requires defence contractors to certify that they will not use Anthropic’s Claude models in work carried out for the armed forces.

But the issue has not been confined to the United States, and concerns have spread across the Atlantic. The European Commission has also initiated a series of meetings with Anthropic to agree on the terms and conditions for future access to advanced models such as Mythos, in order to ensure security. A Commission spokesperson reacted immediately, emphasising that Brussels ‘takes note’ of Anthropic’s decision and that Brussels ‘is assessing’ the matter: “A number of powerful AI models are about to come onto the market. Whilst these models offer significant computational capabilities, they also raise serious cybersecurity concerns,” the spokesperson said.

“The Commission,” added the spokesperson, “will present a comprehensive action plan on cybersecurity and AI. This further underlines the need for Europe to achieve technological sovereignty. It also demonstrates the importance of our existing legislation on cybersecurity and AI.”

The issue was also the focus of talks between the European and US delegations on the sidelines of the G7 summit in Evian, in an attempt to find a long-term solution regarding the sharing of new technological developments in artificial intelligence.

Beyond security concerns, alarm bells have therefore been ringing in Brussels over Europe’s heavy reliance on ‘US-made’ artificial intelligence models and the risk that these might be used as veritable ‘commercial weapons’ by the Trump administration. If Washington decides, as in this case, that its companies must no longer ‘export’ cutting-edge AI innovations, this creates a competitive imbalance between the two sides of the Atlantic that risks severely penalising the Old Continent. The message has come through loud and clear.