Every time Microsoft insists it does not provide technology that enables mass surveillance of civilians, it feels like the universe immediately produces another stack of receipts. The latest batch comes from +972 Magazine, Local Call, and The Guardian. These outlets obtained leaked documents showing that Immigration and Customs Enforcement (ICE) dramatically expanded its use of Microsoft’s cloud and AI services during the first year of the Trump administration’s renewed crackdown on migrants. The scale of the increase is hard to ignore.
ICE’s Use of Microsoft Tools Did Not Just Grow. It Skyrocketed.
According to the leaked files, ICE went from storing roughly 400 terabytes of data on Microsoft’s Azure cloud in July 2025 to nearly 1,400 terabytes by January 2026. That is the equivalent of about 490 million images. This is not a gentle increase. It is a steep, unmistakable spike that cuts directly against Microsoft’s public messaging about responsible use of its technology.
The documents do not spell out exactly what ICE is storing, but they confirm that the agency is using Microsoft’s AI tools, including Azure AI Video Indexer and Azure Vision. These tools can analyze images, read text, and detect faces, emotions, and objects in audio and video files. Civil liberties groups have been warning about these capabilities for years, and the leaked documents show that ICE is using them at a moment when its enforcement operations are more aggressive than ever.
This surge also coincides with a massive budget increase. ICE received an additional 75 billion dollars last July, making it the highest-funded law enforcement body in the country. With that money came a spree of technology purchases, including from Amazon, to support an unprecedented wave of arrests and deportations.
Microsoft’s Principles Start to Look Selective.
When asked whether its anti-surveillance policies apply to U.S. federal agencies, Microsoft repeated its familiar line. The company said it prohibits the use of its technology for the mass surveillance of civilian populations and that it uses internal review mechanisms to assess and address higher risk scenarios.
Here is where things get uncomfortable. The same outlets that reported on the ICE documents previously revealed that Microsoft’s cloud servers were used by Israel’s Unit 8200 to store vast amounts of intelligence on Palestinians in Gaza and the West Bank. That data was then used to plan airstrikes and arrests. Microsoft eventually revoked the unit’s access to Azure. This was the first known instance of a major tech company cutting off an Israeli military entity.
Brad Smith, Microsoft’s president and top public policy executive, framed the decision as a principled stand. He said, “We do not provide technology to facilitate mass surveillance of civilians.” The leaked ICE documents raise an obvious question. If Microsoft’s most senior policy leader can identify and shut down misuse of the company’s services in Israel, why does the company seem unable or unwilling to apply the same scrutiny at home?
Employees and Advocacy Groups Have Been Sounding the Alarm for Years.
This is not Microsoft’s first controversy involving ethically questionable government contracts. During the first Trump administration, employees revolted when they learned the company had contracts with ICE at the same time the agency was separating migrant families at the border. Microsoft CEO Satya Nadella tried to calm the backlash by insisting that Microsoft was only providing legacy mail, calendar, messaging, and document management services, not tools used in enforcement.
Fast forward to today, and the leaked documents show that ICE has significantly expanded its use of Microsoft’s productivity apps and AI tools. The line between legacy support and active participation is looking thinner every day.
Meanwhile, Microsoft has also faced sustained protests over its ties to the Israeli Ministry of Defense. After a joint investigation revealed that Israeli agencies were using Azure to store millions of recordings of Palestinian phone calls, Microsoft disabled specific IMOD subscriptions. Groups like No Azure for Apartheid and CAIR celebrated the move. Microsoft’s internal review confirmed that elements of the reporting were accurate, including IMOD’s use of Azure storage in the Netherlands and its consumption of AI services.
The Pattern Is Impossible to Ignore.
When you zoom out, the through line becomes clear.
- ICE is using Microsoft’s cloud and AI tools during a period of record breaking arrests and deportations.
- IMOD used Microsoft’s cloud to store surveillance data on Palestinians until the company intervened, and that intervention only came after public exposure.
- Microsoft employees have repeatedly protested the company’s willingness to work with agencies involved in human rights violations.
- Advocacy groups continue to pressure the company to cut ties with government entities that use its tools for ethically questionable surveillance.
Microsoft keeps insisting it is guided by principles. Those principles seem to activate only after journalists publish leaked documents, employees stage sit ins, or protesters show up at the company’s own events.
Selling Privacy to Consumers While Selling Surveillance to Governments.
The contradiction at the heart of all this is almost too obvious. Microsoft markets itself to consumers as a privacy protecting, trust first company. At the same time, it sells cloud and AI tools to government agencies whose missions often run counter to those values.
It is hard to take the “your data is safe with us” messaging seriously when the same infrastructure is being used to power deportation dragnets and mass phone call surveillance. It is even harder when the company’s responses to these revelations feel reactive rather than principled.
Microsoft wants to be the company that stands up for privacy. Right now, it keeps finding itself on the wrong side of the people whose rights are most at risk.
