Google workers protest $1.2 billion 'Project Nimbus' contract with Israel: Report

A growing number of Google employees have joined protests calling on the company to drop the $1.2 billion contract, jointly held with Amazon, that provides Israel with cloud and machine learning services, codenamed Project Nimbus. 

According to exclusive coverage of Time magazine, the increasing influence of No Tech For Apartheid, a protest group that has taken a stand against Google's collaboration with Israel and now has around 40 Google staff among its members.

On 4 March, Eddie Hatfield, a 23-year-old Google Cloud software engineer, stood up at Mind the Tech, an annual conference promoting the Israeli tech industry, shouting, "I am a Google Cloud software engineer, and I refuse to build technology that powers genocide, apartheid, or surveillance!"  

Three days later, Google fired Hatfield. 

In a statement to Time magazine, the Google spokesperson said, "This employee disrupted a coworker who was giving a presentation – interfering with an official company-sponsored event. This behavior is not okay, regardless of the issue, and the employee was terminated for violating our policies."

Following Hatfield, two Google employees also recently announced their resignation from the tech giant due to the company's involvement in the $1.2 billion joint effort between Google and Amazon to provide AI and cloud computing services to the Israeli administration and military, according to the Israeli finance ministry, which announced the deal in 2021.

Many workers see Google's firing of Hatfield as an attempt at silencing a growing threat to its business. 

"I think Google fired me because they saw how much traction this movement within Google is gaining. I think they wanted to cause a kind of chilling effect by firing me, to make an example out of me," Hatfield told Time, who agreed to speak on the record for the first time since the 4 March incident.

Project Nimbus reportedly involves Google establishing a secure instance of Google Cloud on Israeli soil. This would allow the Israeli government to perform large-scale data analysis, AI training, database hosting, and other forms of powerful computing using Google's technology with little oversight by the company. Google documents, first reported by the Intercept in 2022, suggest that the Google services on offer to Israel via its Cloud have capabilities such as AI-enabled facial detection, automated image categorization, and object tracking.

The unavailability of contract details raises questions about Google's lack of transparency about what else Project Nimbus entails and the full nature of the company's relationship with Israel.

In a statement, a Google spokesperson said, "We have been very clear that the Nimbus contract is for workloads running on our commercial platform by Israeli government ministries such as finance, healthcare, transportation, and education. Our work is not directed at highly sensitive or classified military workloads relevant to weapons or intelligence services." The spokesperson said all Google Cloud customers must abide by the company's terms of service and acceptable use policy. That policy forbids the use of Google services to violate the legal rights of others, or engage in "violence that can cause death, serious harm, or injury."

No Tech for Apartheid's supporters say that their movement is gaining momentum amid fresh revelations about AI's involvement in Israel's bombing campaign in Gaza and the recent killings of foreign aid workers by the Israeli military.

According to Google workers, they base their protests on three main sources of concern: the Israeli finance ministry's 2021 explicit statement that Nimbus would be used by the ministry of defense; the nature of the services likely available to the Israeli government within Google's cloud; and the apparent inability of Google to monitor what Israel might be doing with its technology.

Under the terms of the contract, Google and Amazon reportedly cannot prevent particular government arms, including the Israeli military, from using their services. They cannot cancel the contract due to public pressure.

Workers worry that Google's powerful AI and cloud computing tools could be used for surveillance, military targeting, or other forms of weaponization. 

Google workers note that for security reasons, tech companies often have very limited insight, if any, into what occurs on the sovereign cloud servers of their government clients. "We don't have a lot of oversight into what cloud customers are doing, for understandable privacy reasons," says Jackie Kay, a research engineer at Google's DeepMind AI lab. "But then what assurance do we have that customers aren't abusing this technology for military purposes?"

Recent reports in the Israeli press indicate that air strikes are being carried out with the support of an AI targeting system; it is not known which cloud provider, if any, provides the computing infrastructure likely required for such a system to run. 

Google DeepMind, the company's AI division, workers fear that the lab's ability to prevent its AI tools from being used for military purposes has been eroded following a restructuring last year. 

Initially, DeepMind vowed that the tech wouldn't be used for military purposes after Google acquired it. However, despite prior assurances, a series of governance shifts now align DeepMind with Google's broader AI principles. Those principles haven't prevented Google from signing lucrative military contracts with the Pentagon and Israel.

One DeepMind team member, who asked not to be named because they were not authorized to speak publicly, said, "I do think this isn't really our decision any more. Google DeepMind produces frontier AI models that are deployed via [Google Cloud's Vertex AI platform] that can then be sold to public-sector and other clients."

One of those clients is Israel.

Google spokesperson did not address specific questions about DeepMind for the Time report.

Other Google workers point to what they know about Google Cloud as a source of concern about Project Nimbus.

The cloud technology the company ordinarily offers its clients includes a tool called AutoML that allows a user to rapidly train a machine learning model using a custom dataset. Three workers interviewed by TIME said that the Israeli government could theoretically use AutoML to build a surveillance or targeting tool.

There is no evidence that Israel has used Google Cloud to build such a tool, although the New York Times recently reported that Israeli soldiers were using the freely available facial recognition feature on Google Photos, along with other non-Google technologies, to identify suspects at checkpoints. 

"Providing powerful technology to an institution that has demonstrated the desire to abuse and weaponize AI for all parts of war is an unethical decision. It's a betrayal of all the engineers that are putting work into Google Cloud," says Gabriel Schubiner, a former researcher at Google.  

Google spokesperson did not address a question asking whether AutoML was provided to Israel under Project Nimbus.

Members of No Tech for Apartheid argue it would be naive to imagine Israel is not using Google's hardware and software for violent purposes.

"Construction of massive local cloud infrastructure within Israel's borders, [the Israeli government] said, is basically to keep information within Israel under their strict security. But essentially we know that means we're giving them free rein to use our technology for whatever they want, and beyond any guidelines that we set," says Mohammad Khatami, a Google software engineer. 

Current and former Google workers also say that they are fearful of speaking up internally against Project Nimbus or in support of Palestinians, due to what some described as fear of retaliation. 

"I know hundreds of people that are opposing what's happening, but there's this fear of losing their jobs, [or] being retaliated against," says Khalek, the worker who resigned in protest against Project Nimbus.

"People are scared." Google's firing of Hatfield, Khalek says, was "direct, clear retaliation… it was a message from Google that we shouldn't be talking about this."

"What Eddie Hatfield did, I think Google wants us to think it was some lone act, which is absolutely not true. The things that Eddie expressed are shared very widely in the company. People are sick of their labour being used for aparthei," Google software engineer Rachel Westrick told Time magazine.

"We're not going to stop," said YouTube software engineer Zelda Montes of No Tech for Apartheid. 

"I can say definitively that this is not something that is just going to die down. It's only going to grow stronger."