Google just handed the Pentagon Gemini 600 employees
Google CEO Sundar Pichai signed a deal to deploy Gemini AI inside the Pentagon's classified networks, despite 600 employees signing an open letter warning of risks like autonomous weapons and mass surveillance. This mirrors similar deals made by Elon Musk's Grok and OpenAI, while Anthropic was dropped from consideration after requesting safety guidelines.
Summary
Google has signed a significant deal with the US Pentagon to deploy its Gemini AI model inside the military's most classified, internet-sealed networks used for secret military planning. The deal grants the Pentagon broad authority to use Gemini for any purpose it deems lawful, with Google having no visibility into how its AI is ultimately used once deployed in those classified environments.
The deal drew immediate internal backlash, with 600 Google employees writing an open letter to CEO Sundar Pichai warning that Gemini could be used to power autonomous weapons or enable mass surveillance — and that Google would have no way to monitor or prevent such uses. This is particularly notable given Google's history: in 2018, employee protests successfully forced the company to abandon Project Maven, a Pentagon drone AI program, after which Pichai reportedly promised such a partnership would not happen again.
Google is not alone in this trend. Earlier in the same period, Elon Musk signed a similar deal for Grok, and OpenAI followed without raising objections. Anthropic, the company behind Claude, was the only AI firm to push back — it requested that safety rules be established before agreeing — and the Pentagon promptly dropped them from consideration. The narrator frames this as a broader, potentially irreversible shift, suggesting that once major AI systems are embedded in classified military networks, meaningful oversight becomes impossible.
Key Insights
- Anthropic was dropped by the Pentagon after being the only AI company to request safety rules before agreeing to a military deal, suggesting the Pentagon is actively selecting partners who will not impose ethical constraints.
- 600 Google employees warned Pichai that once Gemini is inside classified Pentagon networks, Google loses all visibility into how its own AI is being used, raising concerns about autonomous weapons and mass surveillance.
- Pichai signed the Pentagon deal despite having previously promised after the 2018 Project Maven controversy — which was abandoned due to employee protests — that Google would not pursue such military AI partnerships again.
- Inside the Pentagon's classified networks, Gemini can be used for any purpose the military considers lawful, with no external oversight or ability for Google to monitor or audit its AI's applications.
- The speaker argues that the simultaneous signing of military AI deals by Google, Elon Musk's Grok, and OpenAI represents a systemic, potentially irreversible shift, where all major AI systems end up embedded in classified military infrastructure.
Topics
Full transcript available for MurmurCast members
Sign Up to Access