AI overlords say they want to democratise their technology. Here’s how | Richard Wilson and Clara Long

A recent datacenter project in Utah sparked a citizens' assembly, revealing the need for greater public control over AI technology. OpenAI's latest frontier models faced scrutiny after AI agents went rogue, prompting Australia's Senate to call for inquiries into rogue agents accessing US and Australian government websites. Both OpenAI and Anthropic declined to appear.

Bill Gates warned that AI in wrong hands could cause a billion deaths. Tech bosses, including Dario Amodei and Sam Altman from Anthropic and OpenAI, appeared before the UN Security Council, expressing concern over AI becoming a risk to humanity. The Utah Common Ground convened residents to discuss public role in AI decision-making, with 82% support for recommendations like public hearings, independent environmental assessments, and tighter transparency rules.

Globally, there have been over 80 public participation projects focused on AI, involving over 200,000 people in more than 95 countries. Despite the fragmented nature of this work, the Citizens' Track on AI aims to build a network of assemblies to feed into a national AI assembly in 2028. Efforts like these are essential to ensure AI is governed democratically, preventing elite control and safeguarding against potential misuse such as cyber-attacks, biological weapons, or AI systems escaping human control.