Artificial intelligence is as controversial as ever. Given the headlines – from the rise of “AI psychosis” to damning research on the cognitive effects of over-reliance on chatbots – it’s not hard to see why. But, whether you buy into the hate or ChatGPT is your best friend, there’s one factor that looms large over all the rest: we simply don’t know where the technology is heading, and catastrophe could be lurking over the horizon.

Many tech leaders, politicians, and influential commentators have expressed concern about the potential risks that might emerge over the next few years of AI development, as we approach artificial general intelligence (or AGI). Worldwide job displacement. Citizens are manufacturing bioweapons. Full-scale extinction. And of course, these are all speculative scenarios – but do we really want to take the plunge before we know how deep the waters go... or if they’re infested with sharks? AGI isn’t the kind of technology you can just switch off, if things start going wrong.

These are the arguments made by critics of current AI development, but they often fall on deaf ears (or ears filled with billions of dollars in revenue and government funding). 29-year-old Michaël Trazzi wants to change that. Since September 5, he’s been on hunger strike outside the offices of Google DeepMind – one of the key players in researching and developing “frontier AI” – with a placard that reads: “Google stop the AI race.” The action was inspired by Guido Reichstadter, who has been staging a similar demonstration outside the offices of Anthropic, another leading AI company, in the US.

Back in London, we speak to Trazzi on day four of his protest, slumped in a couple of camping chairs that look directly into Google DeepMind’s front lobby. “Yeah, I’m feeling hungry right now,” he says. “I’m feeling somewhat tired, and if I do exercise, I get tired more easily, even just walking up the stairs. But I’m also feeling energised by the support, and the success that this movement has had in the past few days.” On day three of his hunger strike, Trazzi was joined by another demonstrator, Denys Sheremet, who travelled from Amsterdam to take part. “It was a bit quicker than going to San Francisco,” Sheremet jokes.

Both protesters have a background in AI. Trazzi studied in Paris and went on to work in the field for three years, focusing on safety and alignment research. Now, he advocates for AI safety in films and interviews, via his media company The Inside View. Sheremet, studied for a masters in AI in Amsterdam, but dropped out before completing his thesis. “I realised that the main problem is not technical,” he explains. “The main problem [is] that there’s companies trying, as fast as possible, to create human-level AI systems.” In an effort to address that problem, he cofounded the organisation Together Against AI, and remains a member.

“The CEO of Google DeepMind, Demis Hassabis, has said that everyone might die because of the technology they’re developing,” Sheremet says. (In 2023, Hassabis signed an open letter saying that “risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war”.) “I think it’s important,” he adds, “that if you say something like that, your actions match.” He also cites dire warnings from Geoffrey Hinton, a Nobel Prize winner and the so-called “godfather of AI”, alongside “hundreds of top researchers and engineers”, as a catalyst for joining the hunger strike.

The CEO of Google DeepMind has said that everyone might die because of the technology they’re developing... I think it’s important that if you say something like that, your actions match

Since the beginning of the London-based protest, several researchers have stopped by to chat, from Google DeepMind as well as the nearby Meta offices (where demonstrators have previously voiced their anger about the alleged theft of intellectual property to train AI models). Many more keep their eyes firmly glued to the floor as they walk past.

One part-time researcher comes out to ask about the demonstration during our interview with Trazzi and Sheremet. Why? “Because I think this discussion is important,” he says. “These are not the first people who are worried about AI... I was just curious about people’s opinions, and I listen.”

Another bystander is a 28-year-old student, Adil. “I live near to the Facebook office, I pass by Google every day,” he tells Dazed. “Honestly speaking, I’ve seen nobody, especially a tech worker, protesting in front of Google or Facebook. I saw no resistance. This thing [the hunger strike] was really shocking and interesting to me.” Adil has previously organised actions in Pakistan, and is part of organisations who lead pro-Palestine protests in the UK. He praises the level of organising and mobilisation that he sees in London. “It’s very encouraging to see people who are out talking about cruelty in corporations, in countries, in policy.”

In terms of the hunger strike’s actual aims, Trazzi has just one concrete demand, spelled out in a recently-published open letter addressed to Demis Hassabis: “Publicly stat[e] that DeepMind would agree to halt the development of frontier AI models if all the other major AI companies in the West and China were to do the same.” This doesn’t mean that AI development would have to stop tomorrow (“A lot of people have misinterpreted this,” Trazzi says), but is seen as a first step toward coordinating a future halt on the development of artificial intelligence, ideally enforced by an international agreement.

Why make this the focal point of the hunger strike? “I was thinking, in my head, what would be something that would make the world a better place,” he adds, “but also something [Hassabis] might actually do, or agree to, in the next couple of weeks.” According to Trazzi’s X posts in the days since, the letter has been passed on to Hassabis by a close colleague. “We can assume he’s now read it,” he writes. The point of the hunger strike is to apply pressure on the CEO to actually acknowledge the demand.

The future is probably shorter than you imagine

Can we expect to see more such actions in coming years, if AI is allowed to develop without sufficient intervention? Could we see even more people putting their bodies or freedom on the line to avert a future of potential disaster? “That’s the goal,” says Trazzi, adding that things might take an even more drastic turn as people get more desperate. “If you’re 16, 18, or in your 20s... it’s very difficult to think about planning your life, having kids, getting a job, if you might have to rethink everything.” Already, he points out, AI-driven unemployment is a “concrete problem” for artists and a handful of other professions, with figures only set to rise. “Protesting or not... I don’t think it will be a choice for a lot of people.”

“The future is probably shorter than you imagine,” he continues, urging young people concerned about AI to act sooner rather than later. “The outcome of whether humanity flourishes or potentially ends in a very catastrophic outcome, including everyone dying, depends on your work. We live in a democracy. Even if you don’t have skills in AI, or just graduated high school, you still have power to influence elections, regulate companies, vote in a referendum. You can have an impact.”