Google Signs Classified AI Deal with Pentagon, Exits Drone Swarm Contest
Latest News (April 28, 2026)
Google has signed a classified artificial intelligence (AI) deal with the Pentagon, allowing the use of its Gemini AI models for military purposes, despite internal opposition.
The Deal:
- Access: Google agreed to provide API access to its AI systems on classified networks for "any lawful government purpose."
- Existing Relationship: This extends an existing contract between Google and the Pentagon, which already includes Gemini deployment to three million personnel on unclassified systems.
- Safety Measures: The agreement includes language prohibiting domestic mass surveillance and autonomous weapons without human oversight, similar to agreements with OpenAI and Elon Musk's xAI.
Employee Opposition:
On the same day, over 580 Google employees urged CEO Sundar Pichai to refuse such an arrangement, highlighting concerns about AI safety and potential misuse.
Dropping Out of Drone Swarm Contest:
Bloomberg revealed that Google quietly withdrew from a $100 million Pentagon contest for voice-controlled autonomous drone swarms in February due to internal ethics reviews, citing a lack of "resourcing."
Key Takeaways:
- Line Drawing: Google is attempting to draw a line between general-purpose AI access and specific military applications, but the distinction may be blurred on classified networks.
- Employee Discontent: The deal goes against the wishes of many Google employees who advocate for stricter AI safety measures.
- Government Influence: The Pentagon can request adjustments to AI safety settings, potentially undermining Google's own research safeguards.