Google DeepMind (GDM) employees are using collective bargaining as a means of “establishing morals” at the company in response to its “material support for Israel’s genocide” in Gaza, but now face “classic union busting” tactics from management.
In May 2026, DeepMind employees launched a unionisation bid to end use of the firm’s technology by Israel and the US military, seeking to be recognised under the United Tech and Allied Workers (UTAW) union, a national branch of the wider Communication Workers Union (CWU), and Unite the Union.
Computer Weekly has talked with multiple Google DeepMind workers who shared concerns related to the firms’ provision of cloud and artificial intelligence (AI) technologies to the Israeli military, which they say have been used to support a genocide against the Palestinian people.
However, Google so far has not responded well to the union effort, with workers telling Computer Weekly that the company has been trying to shut down or sideline internal opposition to the use of its technologies by Israeli authorities.
“Google seems to be rejecting this golden opportunity to genuinely listen to their own staff, the source of all the company’s value, on the fundamental issue of ethics and how those are lived at the company,” said John Chadfield, a CWU national officer for tech workers.
“Googlers don’t want to work for a defence contractor, and they don’t want to only be heard in a superficial way by management. At Google DeepMind, people came to work on things like protein folding, cures for Alzheimer’s, and the detection of cancer. Circling the drain of the military-industrial complex is against everything they want to be involved in.”
While tech firms have long collaborated with the Israeli government – which has, for some time, been accused of running an apartheid regime on the Palestinian people – workers said there has been an increasing supply of powerful military and surveillance technologies to Israel since October 2023, including drones, artificial intelligence (AI-)powered targeting systems, cloud computing infrastructure, data analytics tools, biometric databases and high-tech weaponry.
In August 2024, for example, in a recording obtained by +972 Magazine, Israeli Defence Force (IDF) colonel Racheli Dembinsky publicly confirmed for the first time that Google’s cloud and AI services – along with those of Amazon and Microsoft – were being used as part of military operations in Gaza.
Workers said that, in the process of GDM essentially transforming itself into a defence contractor for the Pentagon and IDF, it has become more resistant to transparency and oversight.
“There is no way of monitoring what your work is being used for. Google turned into a defence company like Palantir but pretended to not be as bad,” said one worker, who added that at least if you work for a company like BAE Systems, you know exactly what your work will be used for.
“Google just doesn’t tell you what your work is contributing to. If your work is involved in the commission of war crimes, you could be personally complicit and should know about it as an employee.”
Another worker added that the use of Google’s public cloud storage and processing capabilities by the IDF is a necessary precursor to its AI targeting in Gaza, meaning “Google is in the military kill chain” whether its services are used in direct targeting operations or not.
Workers said that while employee-led efforts to oppose the provision of these technologies to the Israeli government were already underway by January 2025 – including internal petitions sent to company management – their concerns became particularly pressing in the wake of Google’s decision to drop its longstanding pledge to not build AI-powered weapons or surveillance tools a month later in February 2025, culminating in the union drive.
“The petitions just weren’t working so we felt we had to go down the union route,” said one worker, who added that without some sort of legal agreement, it was clear that management could change their mind at any time, even if they had agreed to any demands put forward in the petitions. “They can’t ignore us the same way with the union. We see collective bargaining as a route to establishing a set of morals or values at the company.”
According to another worker, “people [at GDM] were in denial” about the supply of cloud and AI technologies to the IDF – and the purposes they were being used for – until the principles were dropped. They described Google’s decision to jettison its longstanding AI principles as a “watershed moment” for the union effort, highlighting that many employees had joined when the principles were in place.
“The mood changed internally. They were building the tech under those guiding principles, which was then repurposed for things the company explicitly said was out of scope,” they said, adding they were also dropped at a time when more and more was becoming public about Google’s supply of technology to the IDF. “Ethics are complicated, but not when it comes to genocide – something has gone terribly off the rails.”
‘Textbook union busting’
However, multiple workers told Computer Weekly that Google DeepMind has not taken the union push well, employing a range of tactics to shut down or at least discourage participation. The workers said management’s behaviour amounts to “classic union busting”.
This includes “third partying the union”, referring to a common anti-union tactic where employers frame the union as an external, interfering “third party” rather than recognising it as the efforts of the workers themselves.
“The union is cast as undemocratic by Deep Mind,” said one worker, who added the firm is attempting to create a “fake separation” between employees and the union.
Workers said Google Deep Mind has also been using its HR department to set up “informal” meetings with employees for breaking its non-solicitation policy, usually for people sending critical messages internally, rather than the formal disciplinary routes available.
According to one worker, this practice is Google DeepMind attempting to avoid formal channels and not put anything in writing, while also attempting to restrict worker organising: “It’s not blatantly illegal, but it’s definitely in a grey area legally.”
In one instance, an AI engineer – who was ultimately dismissed in September 2025 for their further dissemination of flyers and posters internally – was called into a meeting with HR for internally communicating with colleagues about the union effort, where they were “actively discouraged” from making any communications that either criticised the change to Google’s AI principles, or linked the company to Israel’s conduct in Gaza.
“Google fired me for stating the obvious: our work on AI was sold to facilitate genocide,” said the engineer in May 2026, who is not sharing their name at this stage but has made a formal legal claim against the company via a UK employment tribunal.
Despite GDM’s effort to disrupt union organising, however, workers said more and more employees were joining in response to the firms underhanded tactics. As it stands, Computer Weekly has been told that union density is high, with 98% of workers at Google Deep Mind backing the formal request for recognition sent to the company in May 2026.
Workers said while they are pursuing a voluntary recognition agreement with the company (which allows for collective bargaining negotiations on a wider remit of topics and issues), they will push for statutory recognition (which limits negotiations to pay, hours of work and holidays) if needs be.
The employees’ letter gave management 10 working days to voluntarily recognise the CWU and Unite as their representatives, or take other steps such as agreeing to mediated negotiations, before a formal legal process is launched to force recognition.
In response, GDM declined to voluntarily recognise the union, but said it would speak to employees about the matter via conversations with mediators present.
However, Computer Weekly has been told by workers that the company were essentially “filibustering us and not wanting to discuss anything real”, and that no one from management with the “potential power to make any meaningful decisions” was in the room for the mediated talks.
“We asked them to put someone from management in the room otherwise there is no point negotiating, but they declined,” said one worker.
Commenting on Google’s anti-union tactics, Chadfield added: “We said from the start that Alphabet management will hand it off to HR administrators as a box-ticking exercise, and that is exactly what they’ve done. Senior management have refused repeated requests to come to the table, and as a result negotiations have made very little progress.”
Computer Weekly contacted Google DeepMind about every aspect of this story.
“We strongly believe our employees should make an informed decision about whether or not to join a union,” said a spokesperson. “We are still in the early stages of the process, have acted in good faith from the outset and continue to engage constructively in unionisation discussions.”
Previous opposition
The union push at GDM is the latest development in a string of opposition by Google employees to the company’s collaboration with military entities.
For example, Google recently agreed to let the US Department of Defense use its AI models for classified work, a move opposed by more than 600 employees. Google staff worry how the technology will be used given the deal could reportedly open the door to autonomous weapons and mass surveillance of US citizens.
Google employees have also previously protested the ethics of contracts such as Project Nimbus, a joint programme with Amazon to make cloud computing and AI tools available to Israel during its campaign in Gaza.
In July 2025, an investigative report published by Francesca Albanese, the special rapporteur for the “human rights situation in Palestine”, concluded that technology firms globally are actively “aiding and abetting” Israel’s “crimes of apartheid and genocide” against Palestinians.
In particular, she highlighted how the “repression of Palestinians has become progressively automated” by the increasing supply of powerful military and surveillance technologies to Israel, including drones, artificial intelligence-powered targeting systems, cloud computing infrastructure, data analytics tools, biometric databases and high-tech weaponry.
She said that if the companies supplying these technologies had conducted the proper human rights due diligence – including IBM, Microsoft, Alphabet, Amazon and Palantir – they would have divested “long ago” from involvement in Israel’s illegal occupation of Gaza and the West Bank.
“After October 2023, long-standing systems of control, exploitation and dispossession metamorphosed into economic, technological and political infrastructures mobilised to inflict mass violence and immense destruction,” she said. “Entities that previously enabled and profited from Palestinian elimination and erasure within the economy of occupation, instead of disengaging are now involved in the economy of genocide.”
