EU lawmakers rebuke Anthropic for sending junior staffer to AI safety hearing

European lawmakers rebuked Anthropic on Tuesday for sending a junior technical employee to testify before the European Parliament's Committee on the Internal Market and Consumer Protection, as the AI company faced pointed questions about the cybersecurity risks posed by its advanced Claude Mythos model.

The hearing, held on July 14, saw Donny Greenberg — listed on Anthropic's staff as a "Member of Technical Staff" who joined the company in April — appear before the IMCO Committee via video link. MEPs had expected Anthropic's head of public policy to attend and expressed frustration that the company had not sent a senior official to address what they described as pressing concerns about EU security and digital sovereignty.politico+2

Cyber Capabilities Under the Microscope

The exchange of views focused on the safety and cybersecurity risks posed by Anthropic's advanced AI systems, particularly Claude Mythos, which the European Parliament has described as capable of carrying out "complex cybersecurity tasks with little human input, such as finding software weaknesses and completing multi-step actions". The hearing also touched on the application of the EU AI Act and European digital rules to Anthropic's models.europarl.europa+3

The session comes against a backdrop of rising tensions between Brussels and Anthropic. In June, the company revealed it would "suddenly deactivate" its most advanced AI models for all users after a U.S. government directive ordered it to suspend access for foreign nationals on national security grounds. That move prompted the European Commission to examine the "practical consequences" of the decision, and further inflamed concerns about European dependence on American AI firms.reuters

Digital Sovereignty Concerns

MEPs questioned Greenberg on whether the European Union should develop its own alternatives to U.S.-controlled frontier AI systems and whether Anthropic could help Europe reduce its strategic dependence. The Parliament Magazine reported in May that Anthropic had restricted the EU's access to Mythos, "a move that could heighten security risks for Europe".linkedin+1

The hearing follows a broader push by Brussels to address digital sovereignty. The European Commission presented an AI cybersecurity plan in early July aimed at tackling risks posed by cutting-edge AI and reducing reliance on non-EU suppliers. Four EU member states, including Spain, had earlier called for action on the matter.euronews+1

Anthropic has previously called for cooperation on AI models in cyber-defence, but the decision to send a junior staffer to a hearing that had been months in the making left some lawmakers questioning the company's commitment to engaging with European institutions.euperspectives+1