Google Signs Classified AI Deal with U.S. Department of Defense: Internal Backlash Grows Over Permission to Modify Safety Filters

On April 28, 2026, news broke that Google had reversed its past promises and signed a classified AI contract with the U.S. Department of Defense (Pentagon). The deal is sparking ethical controversy as it includes the authority for the U.S. military to request modifications to safety filters applied to Google's core AI models.

The deal, which allows Google's AI models to be used for 'all lawful government purposes,' has reignited internal debates over the ethical boundaries of using artificial intelligence in warfare. Reminiscent of the historic employee revolt in 2018, it suggests a new era of cooperation between Big Tech and national security agencies.

The classified contract reported in late April 2026 paves the way for Google's AI technology to be utilized for a wide range of government purposes. This is interpreted as a follow-up to a series of $200 million contracts the Pentagon signed with major AI labs in 2025. Through this, Google will provide its cutting-edge models within the Department of Defense's classified networks.

This contract represents a significant retreat from the ethical principles Google has long upheld, and it risks allowing AI technology to be integrated into lethal weapon systems.

A key feature of the deal is a clause requiring Google to cooperate in adjusting AI safety settings and filters at the government's request. In contrast to the strict limitations applied to general commercial AI tools, the Department of Defense will have the authority to relax or modify these guardrails for specific military purposes.

Internal Backlash and the Return of Project Maven

Immediate backlash erupted within Google. In a letter, employees pointed out that AI technology tends to centralize power and is prone to errors, urging the company to avoid harmful or unethical deployments. They expressed deep concern, particularly about the risks of deploying AI in the absence of transparency and accountability.

• Potential for civilian casualties due to AI model judgment errors
• Lack of transparent information disclosure regarding military use
• Violation of the company's AI ethics principles and damage to corporate culture

This situation recalls the conflict in 2018 that led Google to abandon 'Project Maven,' a drone-guidance AI contract. At the time, thousands of employees signed petitions and some even resigned to pressure the company, resulting in Google refusing to renew the contract and promising to be cautious in developing AI for military use.

However, the 2026 contract is seen as a complete reversal of that stance. Google has now joined the list of major companies participating in the Pentagon's classified AI projects, alongside OpenAI and xAI. In contrast, Anthropic is reportedly blacklisted by the Department of Defense for refusing requests to remove weapon- and surveillance-related guardrails.

The market reacted positively to the news of the deal. Alphabet (GOOG) shares closed at $348.52 on April 28, up 1.81% from the previous trading day, immediately following the reports. Investors expect the expansion of the Department of Defense's AI budget to become a new revenue stream for Google.

The Pentagon has requested a $2.3 billion budget over the next five years to expand the Maven platform built by Palantir. This signifies a strategic shift toward directly linking battlefield situational analysis AI with weapon systems, suggesting that Google's technology is likely to play a key role in this process.

In conclusion, this 2026 contract demonstrates that the trend of viewing AI as a 'dual-use' technology has become fully established in Silicon Valley. Despite strong protests from internal employees, the strategic alliance between tech giants and the Department of Defense is expected to accelerate further under the banners of national security and technological supremacy.