OpenAI backs mandatory AI safeguards and 'compute campuses' as Democrats near power

A push timed for a new Congress

As technews.tw reported, OpenAI is getting ready for a big policy and public-relations effort in Washington. It starts with the lame-duck session, the stretch after the election when the outgoing Congress is still in office, beginning November 9. The backdrop is an expected Democratic win in the House and a possible one in the Senate.

Chris Lehane, OpenAI's head of global affairs, described the midterms as a gateway from one era into the next. His team has been stockpiling new and old policy ideas since August 2026, so the company can steer the debate instead of reacting to it.

Speed limits for powerful AI

The first proposal is mandatory safeguards for the most capable frontier AI systems, meaning the most advanced models. OpenAI compares them to speed limits and says humans must keep their hands on the wheel. The company plans a more active push for legislation in Congress.

The idea sits between two camps the report names: doomers, who stress catastrophe, and accelerationists, who want to move as fast as possible. OpenAI calls its approach a human-centered third path.

Data centers get a new name

Local resistance to data centers, which use a lot of energy, has been strong across the US. It has already pushed Amazon and Microsoft to promise not to use nondisclosure agreements when they negotiate with local governments.

OpenAI wants to call its facilities 'AI compute campuses'. It argues they can transform regional economies, and it backs state laws giving residents a real say in construction and in how benefits are shared. Its example is a large project in Pike County, Ohio, which it expects to create thousands of jobs.

Workers, allies and 2028

Two more proposals cover worker support during the transition, plus more education and skills training and lower barriers for startups. The fourth calls for the US to lead global AI standards that reflect democratic values.

OpenAI sees the midterms as a preview of the 2028 presidential race. Lehane argues that simply saying AI is dangerous will not satisfy voters, who want a leader who can guide the country through change.