The AI labor fight is already here
Researchers and workers are coming together to help people better understand AI—so they can more effectively fight back against the technology’s forced implementation
Nicole Froio
Real journalists wrote and edited this (not AI). Independent, community-driven journalism survives because you back it. Donate to sustain Prism’s mission and the humans behind it.
When Anna Iovine and other unionized workers at the media company Mashable began the collective bargaining process with their employer in 2024, they had no idea their efforts would become a successful case study in organizing for AI workplace regulations. At the beginning of the process, workers with the Ziff Davis Creators Guild that included Mashable, Lifehacker, and PC Mag were mostly concerned with wages and healthcare.
“The bargaining committee knew we wanted to have some AI guardrails in our next contract, but it wasn’t the focus of the campaign at first,” Iovine told Prism. “But as the campaign went on—we campaigned from February to July 2024—it became more important to people.”
According to Iovine, the associate editor of features at Mashable, two factors fueled workers’ growing concerns about AI implementation at work. First, at the bargaining table, management seemed “very, very resistant” to being transparent about using AI in the Mashable newsroom. Second, the growing list of AI-related media scandals began to weigh heavily on workers. This includes the Sports Illustrated debacle, when the magazine published several fully AI-generated articles and then deleted them when caught by tech-focused publication Futurism. This, and other AI journalism stories, led unionized workers at Mashable to push for protections against AI-related layoffs and salary reductions.
While management initially pushed back on these demands, workers won the collective bargaining agreement four days before they were scheduled to strike. The final contract contains a worker-led definition of generative AI, the required implementation of an AI subcommittee composed of management and workers, the prohibition of AI use to impersonate employees without their consent and without public disclosure, and mandatory public labeling of any AI use.
Mashable is a hopeful example of successful collective bargaining with regard to AI at a time when the technology threatens journalists’ and workers’ livelihoods across industries. The case study is a part of a wider global effort to fight back against forced AI implementation in the workplace. From organizing for better working conditions to teaching people about the harms of AI, the AI labor fight is here.
Is this the future?
When Alex Hanna of the independent organization Distributed AI Research Institute (DAIR) looked for case studies on fighting AI and automation at work, she knew she found one with Mashable.
Hanna, DAIR’s director of research, was in search of case studies for the Luddite Lab Resource Hub, a website that provides resources for unions and organizers fighting AI. The hub features a compilation of successful organizing case studies, along with organizing toolkits, a research and policy resource library, political education materials, and AI primers.
“Luddite Lab starts from the perspective that generative AI specifically is anti-labor,” Hanna told Prism. “And from the perspective that AI is not inevitable, and that we need to build different technological futures and listen to workers who are the closest to the technology they are either being forced to use, or being the ones who are providing the data that is used to train AI.”
As Big Tech continues to impose AI on the workplace, insisting it cuts costs and improves productivity, these same companies are entering uncharted territory while also operating in secrecy, leaving many unknowns about AI’s impact on the economy and labor force. Hanna said these factors impact workers’ autonomy and choice.
Hanna told Prism that Luddite Lab is giving workers the language they need to express concerns about AI at work. This includes its recent primer dispelling common myths about generative AI, including that it makes work easier and gives workers more time for interesting tasks. The reality, according to Luddite Lab, is that generative AI “intensifies work by adding new tasks, increasing workload, and breaking existing work processes.”
As both interest in and backlash against AI intensifies, Hanna points out that the internet is oversaturated with AI discourse and misinformation. The Luddite Lab was designed to cut through the noise so organizers and workers aren’t forced to wade through massive amounts of information, allowing them to organize more effectively. In the Mashable case study, for example, knowledge about AI is highlighted as a winning strategy for the union members.
“AI is not inevitable,” Hanna said. “It is wielded like a cudgel against workers, and workers can control how technology is used at work.”
Making information about Mashable’s organizing strategies available after the union’s victory became very important to Iovine—especially as AI got more popular in 2025 and 2026.
“It is very hard to form a union in the U.S.,” Iovine told Prism. “It is very hard to win, and I think we’re kind of conditioned to not fight for ourselves … I’m hoping the more information that I make public, the more I talk about the campaign, the more people get ideas for their own thing.”
Organizing case studies also provides frameworks for addressing AI in the workplace where none existed before, and they get workers talking about their AI boundaries in a landscape that increasingly treats the technology as an inevitability. Workers do have options for how and when to use AI. For example, in another case study published by Luddite Lab, The New York Times Tech Guild won protections against management’s adoption and use of generative AI and established a member-led AI committee to oversee shifting positions on the technology.
“Some people were like, ‘Well, if we’re going to use it, maybe we need to set our own standards,’” explained Benjamin Harnett, chair of the NYT Tech Guild’s Generative AI committee and a staff software engineer at the Times. “As workers, if we have this tool available to us, what is right from a worker’s perspective? What is right from the judgment of somebody who actually knows how to do this?”
Harnett further emphasized that workers—not management—know best how useful or weak a tool is. “Frankly, most of the folks in leadership are not down in the day-to-day and do not really understand what the work is that we do, how we do it, what is the secret sauce,” he said. “There’s a reason they need folks like us.”
In May, the NewsGuild of New York filed two grievances and an unfair labor practice charge against the Times. In the filing, unionized workers accused management of violating the current contract by using AI programs to surveil them and monitor their performance. Management also allegedly failed to notify the union of its use of AI to track workers’ efficiency.
Prism reached out to The New York Times for comment regarding the filed grievances, but received no response.
According to Harnett, rolling out AI surveillance increased the pressure workers felt to use AI in hopes of increasing their productivity. The software engineer explained that AI programs used by management monitored individual and team “effectiveness.”
“It’s counting the number of things you did and comparing them to some magic industry standard with no indication of where that comes from,” Harnett explained, adding that when the standard isn’t met, management pushes generative AI to increase productivity. “It’s basically a full frontal assault [to force people to] start using this, and show that we’re using it, and now quite a few people are using it,” Harnett said.
Once you realize that there are many more people like you, it reduces the power of that voice that says ‘this is the future.’
For Harnett, it was important to make the AI push from Times management public so that workers in the broader industry would begin talking to each other about AI—and it worked. He said he heard from colleagues that they too were uncomfortable with the use of AI, and they assumed they were the only ones who felt that way. “So sharing that is really important because once you realize that there are many more people like you, it reduces the power of that voice that says ‘this is the future,’” Harnett said.
“Vast power differences”
Spotlighting AI resistance is also a core tenet of the AI Resist List project, a living directory of “imagined radical alternatives” to an AI future. The project centers community, celebrates human agency, honors local context and history, and rejuvenates the planet, according to their website. The project was co-founded by DAIR, tech journalist and author of “Empire of AI” Karen Hao, Refugee Law Lab’s Petra Molnar, and other AI researchers, journalists, and critical scholars.
“What we wanted to do with the AI Resist List is to start a living archive of resistance to AI and the inevitability narratives that private companies and tech bros have been pushing for the last number of years, especially around generative AI,” explained Molnar, who is also a lawyer, anthropologist, and the associate director of Refugee Law Lab. “What we wanted to do is highlight resistance mechanisms that are already happening around the world to also give people a little bit of inspiration as to what they can do.”
Recognizing the imperialist and neocolonialist impetus of AI and Global North tech industries, the AI Resist List gathers resistance projects and movements from all over the world. The list maps global discontent with AI inevitability by highlighting projects such as Slow AI, a group of artists who use zines and community events to reimagine alternative narratives for AI outside of Silicon Valley ideologies. Also featured is Movimiento por un Uruguay Sustentable, a movement that demanded transparency and accountability for a proposed Google data center in Canelones, Uruguay.
“We really wanted it to be as diverse as possible,” Molnar said about the list. “So that it’s not just a creation of a policy document or going to speak at the UN or something that’s very far removed from the average person’s daily experience. We want to focus on individual action, collective action, worker resistance, and creative storytelling.”
The process of putting the list together required navigating ethical issues never considered by large language models (LLMs). For example, Molnar said that all of the listed projects were contacted and vetted, in contrast with AI chatbots that are trained on vast amounts of data that is scraped from the internet without consent of its owners or verification that the information is accurate.
“We didn’t want to assume that somebody doing really amazing transgressive things, for example, in rural Brazil would want to be featured publicly because there’s also security considerations,” Molnar explained. “We’re really ultimately talking about vast power differences between the people who develop and talk about AI and the people who are resisting it. So, we wanted to be extra sure that whoever was profiled in the AI Resist List was comfortable with that and that it was also beneficial to them and their community.”
While projects such as the AI Resist List exist in an increasingly inhospitable online landscape, Molnar said it’s now more important than ever to “turn to one another” instead of working in silos—especially as the cost of living increases alongside the pressure to use AI in the workplace.
Understanding AI
As conversations about AI became more polarized across the political spectrum, activist Michelle MiJung Kim said she felt there were critical gaps in how people were talking about the technology and its harms. Attempting to address information and compassion gaps in the conversation, Kim founded The Courage Studio, which focuses on building “infrastructures of care,” according to the organization’s website. As part of this work, Kim designed the course “AI Strategy Lab for Social Justice Advocates,” a live online course “for change agents who want to understand AI with more groundedness, clarity, and rigor.”
Kim told Prism that she designed the course because of the overwhelming volume of AI discourse she encountered online—both pro and against—and the lack of discussion for social justice advocates about how best to engage the technology in “a way that feels discerning and grounded and not reactive.”
The course teaches people the fundamentals of AI, including its “supply chain” and how it works, Kim explained.
“To really help people expand their knowledge and awareness around how AI arrives to us with all the embedded costs and harms that are not visible on social media,” Kim said. The course also aims to contextualize AI in larger conversations about the extractive practices of the tech industry, particularly with regard to mining in the Global South and the exploitation of data center workers.
“It also helps people to situate themselves in different struggles … even if you’re not a technologist,” said Kim, more broadly arguing that the ethical issues surrounding AI aren’t totally new, but rather a different version of the capitalist, white supremacist extraction that people have organized against for decades.
“A lot of folks that come to the workshop feel they’re powerless and that they are not equipped to push back,” she said. “What I’m trying to let people understand is I know we have the capacity; we have the skills to understand what is actually happening because we have been doing it for a long time.”
The first cohort for the course represented a broad range of people, from anti-AI nonprofit workers to social justice-minded tech workers navigating AI impositions at work. “We take the time to think about our default stance for AI and what our red lines are,” Kim explained. Deciding on boundaries is an important part of the process, according to Kim, because it can lead to harm-reduction strategies in the workplace if and when employers force workers to use AI.
But the threat of AI isn’t just about how and why we use the tool.
In recent months, tech industry leaders have warned of looming mass unemployment that could affect millions of workers.
“One of the main sources of money [to invest in data centers] is going to be by selling people AI that will do the work of workers much cheaper,” said Geoffrey Hinton, the British computer scientist widely known as the “Godfather of AI,” in a November 2025 discussion with Sen. Bernie Sanders. “And so these guys are really betting on AI replacing a lot of workers.”
Hinton also warned that this mass unemployment will result in an economic reshuffle that will leave “millions of workers behind.” DAIR’s Alex Hanna does not agree with that claim, and believes it’s important to offer alternative narratives and case studies for organizing against AI. The technology is a workers’ rights issue, she argued.
“This is a continuation of the workers’ struggle that we need to be attentive to,” Hanna said.
Despite the confidence of tech moguls such as Bill Gates, who said humans will soon not be needed “for most things,” financial analysts continue to warn that the AI bubble will burst in the next two years.
As it stands, the future of AI is still uncertain—especially for workers—but according to experts who spoke to Prism, being informed about the enemy is key to fighting back against whatever comes next.
Editorial Team:
Tina Vasquez, Lead Editor
Carolyn Copeland, Top Editor
Lauren Hardie, Copy Editor
Nicole Froio
