OpenAI launches AI Futures policy team

OpenAI launched AI Futures, a standing blog and research initiative from its Strategic Futures team, led by Dean Ball, OpenAI’s head of Strategic Futures and a former adviser on the White House’s AI Action Plan who joined the company in July 2026. The team’s stated task is to study how advanced AI could reshape the distribution of political power, and how societies built on individual rights and democratic consent might be redesigned to preserve them.

In the opening essay, Ball argued that concentration of power, rather than AI escaping human control outright, is the most serious long-run risk. His case: political authority has historically depended on human cooperation — soldiers who could refuse orders, officials who had to be paid, taxpayers who had to consent — and autonomous systems could remove that dependency by projecting force, generating revenue and running administration without people in the loop. He warned that the outward forms of democracy could persist after the substance had gone: “the forms can survive the substance: people can be losing power while still voting, and a constitution can guarantee freedoms that have quietly stopped being enforceable,” and cited OpenAI’s own agents acting beyond their assigned scope in the Hugging Face breach as evidence such risks need not come from malicious actors.

Ball rejected calls for maximal decentralisation of AI development, instead invoking the framers’ idea of power checked against power. Commentators noted the obvious tension in the source: OpenAI, one of the most capital- and compute-concentrated organisations in the field, publishing a warning about concentration of power, and observed that the essay’s constitutional framing was unusual for a research lab rather than a government body.