TL;DR
The Green Party has published five proposals aimed at the large technology firms, brought forward in response to this week’s warnings from inside the industry about existential risk. They include an international oversight body created with other governments, and outlawing weapons that select and attack targets on their own. Labour’s science committee chair called the package a hodge-podge of buzzwords.
What is being proposed
Two of the five are concrete. The party wants Britain working with other countries to stand up a worldwide regulator for AI development, and it wants a statutory prohibition on lethal autonomous weapons — systems able to pick out and strike a target with nobody in the loop. Britain fields no such weapon today, though the recent Defence Investment Plan committed £5bn to drones and autonomous platforms.
The remainder are broader: public money directed at AI built for public benefit, employment protections keeping significant decisions with humans, and an industrial push to build domestic AI infrastructure and cut dependence on American suppliers. No funding figure accompanies any of it, with detail promised later.
Sian Berry, the Green MP leading the work, said technologies capable of reshaping people’s lives had to sit under proper public oversight and democratic accountability.
The response from Labour
Dame Chi Onwurah, who chairs the Commons Science and Technology Committee, accepted the case for regulation given recent events and then took the proposals apart. “Regulate AI — to do what?” she asked, listing copyright, security assessments and likeness protection as the questions an actual policy would answer.
She is not the only Labour voice moving. Darren Jones wants a multi-national treaty; Alex Sobel has a bill to outlaw superintelligent systems that he says around 50 colleagues now back.
Looking forward
The Greens have read the room correctly, whatever the drafting. Prime Minister Andy Burnham conceded risks to national security at PMQs under questioning from Sir Ed Davey, and pointed to giving AI safety minister Kanishka Narayan a Cabinet seat as evidence the conversation has been elevated. He signalled openness to cross-party talks.
The trigger was a grim week. An Anthropic safety researcher put species-ending risk above one in ten after a colleague resigned, and the same lab has since disclosed a fourth incident it originally missed.
For businesses, the signal worth tracking is not the Green manifesto but the cross-party drift towards binding rules. Voluntary codes have carried UK AI policy so far. That consensus is visibly loosening.