Where is the AI safety movement?

This opinion piece originally appeared in Waging Nonviolence on 5 June 2025. This is part of our new programme of work exploring how social movements can support the safe development and governance of artificial intelligence. If you know of individuals or organisations working in this area - or have ideas for who we should be speaking with, we’d love to hear from you. Please get in touch.

For all the disruption AI is bringing, there is no movement holding it to account. Here are five key ingredients needed for a mobilization to make headway.

Cathy Rogers ≈ 6-min read

Whether or not we realize it, artificial intelligence is reshaping the world with breathtaking speed. Promoted as a panacea to cure diseases, revolutionize industries and tackle the world’s most intractable problems, AI is also quietly displacing workers, amplifying biases and creating surveillance systems pervasive and pernicious enough for any dystopian. As if that weren’t enough, there’s the prospect of super intelligent AI outpacing human control altogether.

So where are the humans in all this? Where are the protests, the people out on the streets, the demands for safeguards, for oversight, for input from normal people? For all the disruption AI is bringing — and we’ve barely scratched the surface so far — there is no movement holding it to account, no mass mobilization demanding that this world-changing technology work for the public good.

There have been some sparks of visible resistance. In Hollywood, the Writers Guild of America and SAG-AFTRA walked off the job in 2023, demanding safeguards against AI-generated scripts and digital replicas of actors’ performances. For the writers and actors, these were not abstract concerns. Studios openly floated the idea of scanning actors’ likenesses for permanent reuse, cutting performers out of the creative process entirely. Yes, their strikes were about wages, but they were also about something more fundamental: preserving human creativity.

Google workers also staged walkouts when the company got involved in Project Maven, a Pentagon initiative to use AI and drones to assist in warfare. Their resistance led Google to back out of the program, demonstrating, at least in this case, the power of coordinated dissent.

One other example of resistance occurred after the wrongful arrest of Robert Williams — a Black man living in Detroit who was falsely identified by a facial recognition system, revealing the tangible human cost of biased algorithms. Williams’s arrest became a rallying point for activists, who forced cities like San Francisco to ban facial recognition technology outright.

These moments matter. But they remain disparate and localized, addressing specific harms rather than the larger systemic risks AI poses. Unlike the climate movement, which transformed fragmented concerns into a global call for action, AI’s sparks have yet to connect and take flame.


At my organization, Social Change Lab, we study social movements — how they start, how they grow, what makes them succeed or fail. Our research can provide insights for a future AI safety movement. We know that social movements don’t emerge in a vacuum. In our view, there are five key ingredients that consistently determine whether a movement will catch fire:

1. Grievances: The most obvious ingredient is anger. People must feel that something is wrong, that there is an injustice both significant and shared. While grievances certainly exist in the case of AI, they remain fragmented and largely under the radar. By and large, people love playing with their new AI tools in the form of ChatGPT, Claude and all the other brain boxes. This is the first issue facing the latent AI safety movement: People are currently more infatuated with AI than fearful of it. AI’s risks often feel too abstract or too distant. Factories don’t become automated overnight, it’s a gradual process; algorithms reinforce inequalities that already exist and we have become habituated to; a future where machines surpass humans is just too hard, literally and emotionally, to imagine. Without human-centered stories, people find it hard to connect.

2. Trigger events: Movements often coalesce around a galvanizing moment, such as a high-profile disaster or injustice, something that puts the risks in the baldest terms. Trigger events don’t so much create new movements as amplify the message of already existing ones. Examples include the Chernobyl disaster, which mobilized the anti-nuclear movement, and George Floyd’s murder, which reignited Black Lives Matter. Unfortunately, trigger events are usually awful things. Although AI has not yet had such a moment, it could look something like an autonomous weapon going disastrously wrong, an AI system making a catastrophic decision in a high-stakes setting like health care, an economic crisis caused by sudden mass unemployment or a data breach exposing millions to harm.

3. Leaders and narratives: Strong charismatic leaders and compelling stories often help join scattered fragments of grievances into a single collective narrative. The civil rights movement had Martin Luther King Jr. and the climate movement has Greta Thunberg. AI safety has yet to find its lead vocalist. When Geoffrey Hinton, the so-called godfather of AI said he’s worried about the end of humanity, you might think that would do it. But that news cycle moved on. Other candidates, like Timnit Gebru, the former Google researcher who exposed the biases in AI systems, faced corporate retaliation for her work. We need credible, passionate people who can tell us about AI threats in human terms and move us to act.

4. Resources: Movements need funding, networks, and media attention to scale and sustain over time. Activism generally is consistently under-supported and under-resourced. Just as the climate movement is up against the immense budgets and power of the fossil fuel industry, any AI safety movement will face the goliath of the tech industry driving AI development.

5. Political opportunities: Movements flourish when systems are open to change. Growing regulatory pressure on Big Tech, at least in some countries, provides an opening, but the entrenched influence of industry lobbying is a formidable obstacle — and AI is complicated because the scale of the potential gains it might bring are so alluring. AI is developing at a pace that outstrips public awareness. We are barely getting our heads around the latest technology while the next 10 iterations of it are queueing up behind. Many people don’t realize how quickly AI technologies are being integrated into hiring, policing, health care and governance. By the time the risks are fully visible, it may be too late to act.

Movements can make headway when these ingredients are in place. Consider the climate movement, which took decades of building grassroots activism and increasingly desperate alarm bells sounding by scientists before a defining moment arrived. Greta Thunberg’s solitary protest outside the Swedish parliament in 2018 grew into Fridays for Future, inspiring millions of students worldwide to strike for the planet. Extinction Rebellion’s 2019 Summer of Uprising disrupted central London, forcing climate breakdown into the headlines and the language of the “climate emergency” to be adopted by the U.K. Parliament.

The window for mobilization is narrow. Yet history shows that movements can snowball with startling speed when the conditions are right. If a single teenager with a placard can ignite a movement of school children and students around the world, we know it is possible. An AI safety movement must similarly tie the local to the global. It must tell stories that force people to make the connections — link the Hollywood writers’ strike to tech worker walkouts, link wrongful arrests caused by biased algorithms to insiders warning us of existential risks to humanity. And it must move fast.

An AI safety movement needn’t mean a complete rejection of AI — in fact, some prominent social movement scholars argue that it’s crucial for movements to use AI technologies to organize and mobilize. Just as climate activists aren’t anti-energy, but want clean, cheap, sustainable energy, an AI safety movement would focus on better aligned AI. It would demand that AI development be guided by human values: transparency, fairness, accountability and human needs. It would call for people and governments, not corporations, to set the rules.

The stakes couldn’t be higher. AI isn’t just another innovation — it’s a force capable of reshaping humanity. The question isn’t whether it will face resistance. The question is whether the resistance will arrive in time.

---

Cathy Rogers PhD is Senior Researcher at Social Change Lab, a nonprofit that conducts research on protest and people-powered movements to understand their role in social change.

Image at top by ufcw770: WGA Strike 6.21.2023 024. licensed under the Creative Commons Attribution 2.0 Generic license.