TechFlow News, July 14, according to AXIOS, Google DeepMind Co-founder and CEO Demis Hassabis called on the US to establish a new AI regulatory agency, which should have the authority to review the world's most advanced AI models and coordinate industry-wide slowdowns when risks escalate. Hassabis stated that it is time to adopt a more "systematic" approach to AI regulation, funded by the industry, composed of world-class technical experts, and accountable to the US government. He said that the cybersecurity risks currently posed by AI are just "warning signals."
He believes that within the next 18 months, these capabilities as well as more serious biological and nuclear security risks could all appear in open-source models that no government can control. He also emphasized that risks not only come from open-source models, but will also come from more powerful proprietary models launched by major AI labs in the future. It is reported that Hassabis has been quietly promoting this plan for several months, and before the public release, he has communicated with the Trump administration, heads of other AI labs, and European officials, stating that "the feedback heard has been very positive." (Jin10)
