8legs: Eu ai office

#1 out of 1

Anthropic’s AI hacking tech exposes EU AI Office weaknesses

EU AI Office lacks access to Anthropic’s Mythos model and the experts needed to address a potential cybersecurity crisis.
Critics say the AI Office is under-resourced and not sufficiently integrated within the Commission to respond quickly to threats.
The report compares the EU with the U.K., where the AI Security Institute has stronger links to industry and regulatory tools.
EU officials say the AI Office has ‘state-of-the-art model evaluation capacity’ but needs more staff in safety units and technical experts.
EU lawmakers urge expanded staffing and budget to match frontier AI’s scale and speed.
The eight AI safety groups advocate for expanding the safety unit to roughly 160 members by 2030.
EU plans include hiring several staff by June to bolster safety, regulation, and compliance roles.
EU officials acknowledge the AI Office is less than two years old and faces internal power and hierarchy challenges.
The EU Act regulator role includes potential fines up to €35 million, underscoring the regulator’s stake in enforcement.
The report cites UK achievements in AI security analysis as a model for Brussels' upgrade.

Vote 0