Google worried Israeli contract could enable human rights violations

San Francisco: In May 2021, Google announced it had agreed to participate in a $1.2 billion cloud computing contract with the Israeli government and military, saying it was "delighted to have been chosen to help digitally transform" the country.

But four months earlier, officials at the company had worried that signing the deal, called Project Nimbus, would harm its reputation, according to documents prepared for executives that were reviewed by The New York Times.

Google's lawyers, policy team employees and outside consultants -- who were asked to assess the risks of the agreement -- wrote that since "sensitive customers" like Israel's Ministry of Defense and the Israeli Security Agency were included in the contract, "Google Cloud services could be used for, or linked to, the facilitation of human rights violations, including Israeli activity in the West Bank."

The files, which have not been previously reported on, showed that despite Google's public defense of Nimbus over the last three years, the company once had concerns about the contract similar to those of some employees, who have argued that it pulled Google into a long conflict between Israelis and Palestinians.

The documents also offer fresh insight into how the tech giant assessed a contract heralded as a gateway to the Israeli cloud computing market. Though the deal, for seven years, was tiny for a company with $258 billion in sales in 2021, it was an important government contract for Google's cloud computing business, which was struggling to compete with much larger cloud businesses at Amazon and Microsoft. (Amazon also supplies computing services to Israel under the Nimbus deal.)

Google provided Israel with the processing power needed to run applications and artificial intelligence tools, the documents showed, including technology that analyzes images and videos to detect objects. The company also supplied services to store and analyze large amounts of data, along with more mundane software like Google's videoconferencing system.

The contract was a boon for Thomas Kurian, Google Cloud's CEO, who took the helm of Google Cloud in 2019. And while the Nimbus deal was contested at the time, it was a precursor to Silicon Valley's increasingly enthusiastic pursuit of military and intelligence customers. In 2022, the Pentagon awarded its $9 billion Joint Warfighting Cloud Capability contract to Google, Amazon, Microsoft and Oracle.

But the Nimbus deal has been a lightning rod for arguments inside Google, especially since the start of the war in the Gaza Strip last year. Some employees claim Google's technology could be playing a role in the conflict. The company has denied that, saying Nimbus "is not directed at highly sensitive, classified or military workloads relevant to weapons or intelligence services."

"We have been very clear that the Nimbus contract is for workloads running on our commercial cloud by Israeli government ministries, who agree to comply with our terms of service and acceptable use policy," a Google spokesperson said in a statement. An Amazon spokesperson declined to comment.

A spokesperson for Israel's Ministry of Finance said in a statement that Nimbus would help Israel solidify its position as a leading technology hub and improve the day-to-day lives of Israelis, and that the use of AI and machine learning would fuel new technologies and startups that people worldwide would enjoy.

She declined to comment on how the Israeli military was using the technology.

Israeli and Palestinian supporters have been engaged in bitter debates on Google's internal forums. In April, some employees staged sit-ins at two Google offices, criticizing Nimbus on company whiteboards and posted signs. Police arrested nine of the protesters, and Google ultimately fired around 50 workers for participating in protests.

In October, the company barred employees from writing unauthorized messages on whiteboards and displaying signs around the office, according to an internal message viewed by the Times. Signs, posters and materials that do not promote Google-sponsored events and initiatives, or do not contribute to a "safe, productive and inclusive environment," will be removed and may result in "corrective action" for employees, the company said.

Roberto González, a San Jose State University professor who has written about the military businesses of tech companies, said that in the last three years, big tech companies had grown increasingly comfortable with military and intelligence work.

Recently, pushback against these types of contracts has "really come to the fore with the war in Gaza and with the criticisms of the Israeli military in particular on issues around human rights," González said.

Google has a history of employee activism against what workers believe could be the militarization of their technology. In 2018, Google workers protested Project Maven, an agreement to help the Pentagon identify people in drone videos. Google shut down the effort the next year when the contract expired.

The worker opposition to Maven prompted Google to create principles about how it would deploy its AI. It ruled out using its technology for weapons, surveillance or human rights violations.

Three months before Google signed the Nimbus contract, the company's consultants suggested that it prohibit the sale and use of its AI and machine learning tools to Israel's military and other sensitive customers.

Google had taken that approach in other countries but ultimately did not in Israel, according to the documents. The Google spokesperson said the company was "proud to have a large number of public sector cloud customers around the world" and followed "a consistent process for reviewing and entering these contracts," including compliance with its AI principles and other policies.

The company also worried that it would be forced to accept "onerous" risks, such as the possibility that it could run into conflicts with foreign or international authorities if they sought Israeli data and that it might have to "breach international legal orders" under the deal terms, according to the documents.

While it was considering the Nimbus deal, Google engaged its primary human rights consulting firm, Business for Social Responsibility.

Besides recommending that Google not provide AI to the Israeli military, BSR consultants worried that Google would have little understanding of how Nimbus customers in Israel were using its technology. It recommended that Google perform "due diligence" to ensure the services were being used as intended.

Finally, the consultants recommended that Google incorporate its AI principles into the contract, which would commit Israel not to use Nimbus for surveillance or weapons or to harm people, according to the documents.

But when Google negotiated next with the Israeli government, it did not get everything it asked for. The government did not add the AI principles to the contract. But it did say Google had the right to suspend customers if they violated the company's terms of service and acceptable use policy, which forbids clients to use technology to undermine individuals' legal rights, break the law or spread computer viruses, the documents showed.

Under the terms of the deal, Google expected to get the largest share of money from Israel's Ministry of Defense, an estimated $525 million from 2021 to 2028, which dwarfed the $208 million it expected to receive from the rest of the country's central government.

The company anticipated total revenue of $1.26 billion over seven years, including business from Israeli local governments and some of the country's health care providers, the documents showed.

It was a tiny amount for a giant company, but it gave Google credibility with military and intelligence customers that workers had opposed.

Google demonstrated to these customers that it was "open for business," González said. "Employee concerns or protests are not going to stand in the way of the company doing these deals."