700,000 Tech Workers Urge Amazon, Google, Microsoft to Reject Pentagon Military AI Demands

(UNITED STATES) — Worker organizations and unions representing approximately 700,000 employees urged Amazon, Google and Microsoft to reject Pentagon demands for unrestricted AI and defense collaboration, widening a fight over military AI, corporate ethics and government contracting.

The coalition pressed company leaders to set limits on military AI work, including uses tied to autonomous weapons and mass surveillance, and to disclose more about sensitive government contracts, including those linked to immigration enforcement.

Groups tied to employees at major cloud and artificial intelligence providers framed the demands as a response to government pressure around AI and defense work, as executives across the sector pursue more federal business.

The push drew added attention because it reached beyond the Pentagon to call for transparency involving the Department of Homeland Security and its components, including Customs and Border Protection and Immigration and Customs Enforcement.

A recent standoff involving AI firm Anthropic provided an immediate backdrop for the campaign and escalated an already heated debate inside Silicon Valley about safeguards for weapons-related and surveillance use cases.

The Financial Times reported that workers at Amazon, Google and Microsoft want their companies to back Anthropic’s refusal to loosen certain safeguards that could enable autonomous weapons or broad domestic surveillance.

On February 27, 2026, President Trump ordered federal agencies to cease using Anthropic’s “Claude” AI after the company refused Pentagon demands to allow “any lawful use” of its models, including mass domestic surveillance and autonomous weapons.

That dispute sharpened tensions between executives seeking government AI revenue and employees concerned about downstream harms and reputational risk, as worker groups cast the question as one of corporate ethics as well as product governance.

The coalition listed the Alphabet Workers Union, Amazon Employees for Climate Justice, No Tech for Apartheid, and No Azure for Apartheid as part of the organizing effort, tying its message to workers across the three companies.

The campaign’s immigration-related focus landed on the technology that can support enforcement work, with labor and activist groups calling for more disclosure around contracts and services involving ICE and CBP.

Worker groups argued that procurement decisions and vendor relationships can determine how commercial tools get integrated into enforcement operations, including data pipelines, system integrations, and contractor-run services that support tracking and case routing.

DHS reported that its use of AI tools for ICE and CBP has expanded significantly, with tools like “ImmigrationOS” being used to track and route individuals through the enforcement process.

Secretary of Homeland Security Kristi Noem defended the department’s posture during an oversight hearing before the Senate Judiciary Committee on March 3, 2026, addressing criticism of immigration enforcement.

“I want to address the dangerous environment that our ICE officers face on the streets today. They are facing a serious and escalating threat as a result of deliberate mischaracterizations of their heroic work and rhetoric that demonizes our law enforcement,” Noem said.

DHS also reaffirmed its adherence to Directive 139-08, which prohibits using AI outputs as the sole basis for law enforcement actions, while its latest AI inventory showed a 40% increase in AI use cases.

In late January 2026, DHS increased the “exit bonus” for self-deportation through the CBP Home app from $1,000 to $2,600 plus a free flight, reporting that 2.2 million people have voluntarily self-deported since early 2025.

USCIS placed its own compliance warning alongside the current hiring climate in tech, announcing on March 2, 2026 that the FY 2027 H-1B registration window runs March 4 – March 19, 2026 and highlighting increased enforcement under “Project Firewall.”

“Through Project Firewall, the Department of Labor (DOL) in coordination with federal partners including USCIS intends to investigate more aggressively whether employers are complying with their H-1B obligations. Employers will be subject to a higher risk of site visits and investigations by DHS and DOL in 2026 and beyond,” USCIS said.

A new mandatory background requirement for H-1B petitions takes effect April 1, 2026, adding another layer of scrutiny for employers that sponsor foreign workers even as contracting choices shape what projects expand or shrink.

In the Pentagon, Secretary Pete Hegseth has pushed for what was described as “patriotic” tech partners who will not impose ethical “red lines” on military operations, as procurement ambitions collide with employee dissent.

Under Secretary Emil Michael criticized Anthropic’s restrictions in a podcast on March 6, 2026, calling them an “irrational obstacle” and adding: “I need a reliable, steady partner. who’s not going to wig out in the middle” of a conflict.

On March 2, 2026, Congresswoman Robin Kelly sent a formal inquiry to tech CEOs seeking data on DHS subpoenas used to identify individuals criticizing immigration policy, adding pressure on companies already facing internal demands for transparency.

For international students and skilled visa holders inside the tech sector, the dispute has tied workplace choices to compliance risk, as sensitive contracts can change hiring priorities toward cloud infrastructure, cybersecurity, compliance, data systems and AI operations.

Worker leaders have signaled they plan to keep pressing for disclosures about Pentagon, DHS, CBP and ICE-related work, while companies weigh procurement strategy, workforce pressure and federal expectations in a market that increasingly treats military AI as a core line of business.

More details on the agency statements cited by the worker groups and their critics appear in the USCIS newsroom, DHS’s responsible use of AI materials, and Kelly’s oversight records.