Google Drops Pledge: AI for Weapons Now OK

The company that once refused military AI contracts just erased its ethical red lines.

TL;DR: Google quietly removed its commitment not to build AI for weapons or surveillance. The 2018 "AI Principles" that once prohibited AI "for use in weapons" or surveillance are gone. The company that sparked employee walkouts over Project Maven (a Pentagon drone targeting program) now openly embraces military AI. DeepMind's acquisition terms supposedly barred military use. That's over. Google says "democratic nations must serve government and national security needs." Amnesty International called it "shameful." Human Rights Watch warned it could "fuel surveillance and lethal killing systems at vast scale."

What Changed

In February 2025, Google updated its "Responsible AI Principles" page. The changes were stark [1][2]:

What Was Removed

  • Commitment to not develop AI "for use in weapons"
  • Prohibition on AI where "primary purpose is surveillance"
  • Promise to not design AI that "contravenes widely accepted principles of international law and human rights"
  • Pledge to avoid AI that causes "overall harm"

What Was Added

A new justification: "Google is updating its AI principles because the technology has become far more widespread, and companies in democratic nations must serve government and national security needs" [3].

Translation: Ethics are nice, but there's money in military contracts.

How We Got Here

2018: The Project Maven Revolt

In 2018, Google employees discovered the company was secretly working on Project Maven, a Pentagon program to use AI for analyzing drone footage and identifying targets [4].

Thousands of employees signed a letter demanding Google withdraw. Some resigned. The backlash was severe enough that Google pulled out of Project Maven and published AI Principles explicitly banning weapons development.

Those principles weren't optional. They were meant to be binding commitments. Google promised specific things it would never do.

2015: The DeepMind Promise

When Google acquired DeepMind in 2014, CEO Demis Hassabis extracted specific protections. In a 2015 Wired interview, Hassabis said the acquisition terms included that "DeepMind technology would never be used for military or surveillance purposes" [5].

That promise is now void.

2024: The Slide Begins

Google signed a $200 million contract with the Pentagon to provide AI to 3 million military users. The company argued this was "productivity tools," not weapons. Critics saw it as the start of a pivot.

2025: The Mask Drops

With the February 2025 update, Google stopped pretending. The prohibitions on weapons and surveillance were simply deleted.

Google's Justification

In a blog post, DeepMind CEO Demis Hassabis and Google AI head James Manyika explained the change [3]:

"The technology has become far more widespread, and companies in democratic nations must serve government and national security needs."

This argument has several problems:

  • "Democratic nations": Google operates in countries with authoritarian tendencies. Where's the line?
  • "Must serve": No one is forcing Google to build weapons. This is a choice.
  • "National security needs": This phrase has justified surveillance of journalists, activists, and entire populations.

The real driver is competition. Microsoft has Copilot contracts across the military. Amazon has CIA cloud deals. Google doesn't want to be left out.

The Response

Amnesty International

Matt Mahmoudi, an Amnesty International researcher, called the decision "shameful" [6]:

"AI-powered technologies could fuel surveillance and lethal killing systems at a vast scale, potentially leading to mass violations and infringing on the fundamental right to privacy."

Human Rights Watch

Human Rights Watch noted the shift from "refusing to build AI for weapons to stating an intent to create AI that supports national security ventures is stark" [6].

They warned that militaries are increasingly using AI in war, "where their reliance on incomplete or faulty data and flawed calculations increases the risk of civilian harm."

Former Employees

Several former Google employees who left over Project Maven expressed dismay. Their sacrifices (resignations, career disruption, public advocacy) had been for principles that Google discarded when convenient.

What This Enables

With the ethical restrictions gone, Google can now pursue:

  • Autonomous weapons systems: AI that identifies and engages targets without human intervention
  • Mass surveillance: AI-powered monitoring of populations at scale
  • Predictive targeting: Systems that determine who to kill based on behavioral patterns
  • Intelligence analysis: Processing drone footage, satellite imagery, and intercepted communications
  • Cyber weapons: AI that identifies and exploits vulnerabilities

Google claims decisions will be made "case by case." But without binding prohibitions, "case by case" means "yes, eventually."

Part of a Larger Pattern

Google isn't alone. The entire tech industry is abandoning AI ethics commitments:

  • Microsoft: Deep Pentagon integration, Copilot for military
  • Amazon: AWS hosts CIA, military cloud services
  • Palantir: Built explicitly for military/intelligence AI
  • OpenAI: Dropped ban on military use, partnering with defense contractors

The AI ethics movement of 2018-2020 is dead. The companies that once pledged restraint are now racing to arm governments.

What This Means for You

Google's AI powers products used by billions:

  • Gmail, Google Docs, Drive (your documents)
  • Android (your phone)
  • Google Photos (your face)
  • Chrome (your browsing)
  • YouTube (your interests)
  • Google Maps (your location)

The same company now building military AI has unprecedented access to your data. The wall between consumer products and military applications has collapsed.

Google says it won't share user data with military systems. But that promise comes from a company that just deleted its weapons prohibition. Trust accordingly.

What You Can Do

Reduce Dependence

Diversify away from Google products where possible. Use privacy-focused alternatives for search (DuckDuckGo), email (ProtonMail), and browsers (Firefox).

Use Privacy Settings

If you stay on Google services, enable every privacy setting. Turn off activity tracking. Delete data regularly. Opt out of personalization.

Support Oversight

Back organizations pushing for AI accountability: EFF, Amnesty International, Human Rights Watch, Campaign to Stop Killer Robots.

Pay Attention

Corporate ethics pledges are worthless without enforcement. When companies make promises, remember: they can revoke them anytime.

References

  1. Washington Post: Google drops pledge not to use AI for weapons or surveillance (February 2025)
  2. CNN: Google removes restrictions on military AI development (February 2025)
  3. Computer Weekly: Google drops pledge not to develop AI weapons (February 2025)
  4. The Conversation: Google has dropped its promise not to use AI for weapons (February 2025)
  5. Al Jazeera: Google drops pledge not to use AI for weapons, surveillance (February 2025)
  6. Human Rights Watch: Google Announces Willingness to Develop AI for Weapons (February 2025)