AXIOS reports DeepMind co-founder and CEO Demis Hassabis called for a US-led
global AI regulator empowered to review the world’s most advanced models and to
coordinate industry-wide slowdowns if risks escalate. He said the agency should
be industry-funded, staffed by top technical experts, and accountable to the US
government. Hassabis warned current cyber risks are an early signal and that
within 18 months similar capabilities — and more severe biological and
nuclear-security risks — could emerge in open-source models beyond any
government’s control; he added these risks will also come from future
proprietary models. He has briefed the Trump administration, other AI lab chiefs
and European officials and said feedback has been positive.
