Political figures as far apart as the independent senator Bernie Sanders and the former Trump adviser Steve Bannon have jointly warned that artificial intelligence (AI) must not be allowed to run wild.
The two spoke back to back at the Pro-Human Assembly 2026 in Washington, hosted by the Future of Life Institute, a non-profit that campaigns on the risks posed by advanced technology.
Billed by organisers as the first of its kind, the gathering drew about 200 people, according to the Christian Science Monitor.
Attendees ranged from lawmakers and tech experts to parents and faith leaders, united by the view that AI should serve humans rather than the other way around.
"I don't think you could find any two harder partisans than Bernie Sanders on the left, and [myself] on the right," Bannon said, describing the moment as a hinge in history.
The assembly followed the resignation a week earlier of Jacob Coxon, a researcher at the AI company Anthropic, who quit warning that the technology could destroy humankind within a few years and accusing firms of prioritising competition over safety.
Agreement, up to a point
Speakers put forward a range of measures, from requiring AI companies to be more transparent, to protecting whistleblowers, to an immediate pause on the most advanced models.
They agreed that technology firms could not be left to regulate themselves and that a blind race to build ever more powerful systems carried real danger.
Their proposed remedies, however, diverged sharply.
Sanders and the Texas Democrat Greg Casar want Congress to ban superintelligence, a theoretical form of AI that would surpass human ability across most tasks.
The Texas Republican Chip Roy called instead for light-touch guardrails shaped by public opinion, while Bannon wants to rely on presidential action to force a slowdown.
A rare bipartisan moment
Such cross-party alignment is unusual, with Democrats generally pushing for tighter regulation and Republicans stressing the need to stay competitive with China.
The warnings were underscored by a United Nations report in July, which said the gap between rapidly improving AI and effective ways to manage its risks could lead to catastrophic outcomes.
Whether that shared alarm can be turned into law remains uncertain, given the competing definitions of when AI use tips from useful to harmful.
Around 350 words. Three things I dropped because they aren't in the CSM article you linked: the Marriott ballroom detail; Sanders' proposal to take large ownership stakes in AI firms; and the "federal preemption" and "labour voices" framing. If any of those come from a separate source, send it over and I'll fold them back in. Also worth noting the piece was written by Caitlin Babcock, if you want a named byline credit rather than just "the Monitor".
