4 weeks ago
Workers processing Google’s artificial intelligence products successful nan UK person voted to unionize, successful portion retired of concerns astir a woody betwixt nan institution and nan US subject that was announced past week.
In a missive slated to spell to guidance connected Tuesday and shared exclusively pinch nan Guardian, workers astatine Google DeepMind, nan company’s AI investigation laboratory, requested nickname of nan Communication Workers Union and Unite nan Union arsenic associated representatives of nan lab’s UK-based staff.
DeepMind’s UK workers voted to unionize successful April. One of nan workers said they were peculiarly driven by reports that Google was adjacent to reaching a woody pinch nan defense section and pointed to nan US’s “capricious Iran war” and nan Trump administration’s feud pinch Anthropic arsenic indications that nan section is “not a responsible partner”. The woody was yet announced connected Friday.
“I person joined nan national owed to concerns astir AI being utilized to empower authoritarianism, whether done subject aliases surveillance applications, some overseas and domestic,” added nan worker, who requested anonymity because of fearfulness of retaliation. “By unionizing, we are taking nan accepted way for workers to shape and person a say.”
Another worker, who besides requested anonymity, said that galore astatine nan institution had struggled pinch what they had travel to position arsenic their complicity successful Israel’s warfare successful Gaza. The institution provided nan Israeli subject pinch accrued entree to its AI devices from nan early days of nan warfare successful Gaza, nan Washington Post reported past year, and successful 2021, it signed, on pinch Amazon, a $1.2bn cloud-computing statement pinch nan Israeli government.
“Our exertion helped nan IDF,” said nan 2nd UK worker, referring to Israel’s military. “I want AI to use humanity, not to facilitate a genocide.”
Google did not respond to a petition for comment.
Concerns by Google workers and investors person been mounting for years but person peculiarly escalated aft nan institution past twelvemonth dropped a promise not to create militarized AI. That improvement was a driving information for Google DeepMind workers’ national successful nan UK, 2 of them said. While mini groups of Google labor person unionized successful nan US before, nan UK workers are nan first successful a “frontier” AI laboratory to activity national recognition, they said. Google DeepMind is headquartered successful London but has astir a twelve offices crossed North America and Europe. At slightest 1,000 workers will beryllium represented if nan institution recognizes nan union, according to national officials.
On Friday, nan Pentagon confirmed it had reached agreements pinch 7 starring AI companies, Google among them. Others included SpaceX, OpenAI, Nvidia, Reflection, Microsoft and Amazon Web Services. Anthropic, whose exertion is successful wide usage by nan US subject but which has sparred pinch nan Pentagon complete early contracts, was notably absent from nan group.
“These agreements accelerate nan translator toward establishing nan United States subject arsenic an AI-first fighting unit and will fortify our warfighters’ expertise to support determination superiority crossed each domains of warfare,” nan defense section officials said successful a statement.
The Trump management has pushed AI companies to make their devices disposable connected classified networks without nan modular restrictions they use to users. Google’s statement pinch nan Pentagon reportedly includes connection stating: “The parties work together that nan AI System is not intended for, and should not beryllium utilized for, home wide surveillance aliases autonomous weapons (including target selection) without due quality oversight and control.” But that connection is non-binding, and nan statement besides says Google has nary correct to power aliases veto “lawful” authorities operational decision-making.
Workers who voted to subordinate nan national said they did truthful to raise unit connected Google to meet demands already made by different labor astatine nan company, including that it perpetrate not to create exertion “whose superior intent is to origin harm aliases wounded to people”, found an independent morals oversight body, and assistance workers nan individual correct to garbage to lend to projects connected civilized grounds. Should nan institution refuse, they said, they are considering protests and “research strikes”, during which unit abstain from activity expected to importantly amended halfway products specified arsenic Gemini, Google’s AI bot, while avoiding discovery by continuing to execute little important updates.
Workers crossed Google person been progressively vocal astir their guidance to militarized applications of their technology. Last week, amid reports of nan pending deal, much than 600 Google labor signed an open letter to CEO, Sundar Pichai, demanding nan institution not make its AI systems disposable for classified use.
“We want to spot AI use humanity; not to spot it being utilized successful inhumane aliases highly harmful ways,” they wrote. “Making nan incorrect telephone correct now would origin irreparable harm to Google’s reputation, business, and domiciled successful nan world.”
Tech workers person progressively challenged guidance complete nan usage of nan exertion they person helped develop. In 2024, Google fired 50 workers who had protested against Project Nimbus, nan 2021 statement pinch nan Israeli government. At Microsoft, which nan Guardian revealed supplied Israel pinch unreality retention utilized successful nan wide surveillance of Palestinians, workers occupied a institution field pinch signs reference “No Labor for Genocide”. (The institution terminated nan Israeli military’s entree to that exertion aft nan Guardian’s reporting.)
Investors person besides raised concerns. A conjugation of shareholders who ain astir $2.2bn of Alphabet’s shares wrote a letter to Google’s genitor institution past week demanding a gathering and greater transparency astir Google Cloud and AI deployments successful “high-risk” contexts. They cited concerns astir nan institution providing services to US migration authorities, arsenic good arsenic Project Nimbus, and raised questions astir “the effectiveness of argumentation guardrails, soul escalation processes, and Board oversight of AI deployments successful conflict-affected aliases security-sensitive environments”.
In 2018, Google besides dealt pinch wide worker protests complete a subject statement known arsenic Project Maven, successful which nan institution agreed to build AI products for nan Pentagon’s study of drone footage. In consequence to nan backlash, nan institution did not renew nan statement successful 2019 and published a group of principles for its activity connected AI that included nan pledge, now dropped, not to creation AI for weapons. Palantir took complete Project Maven, which continues today.
