Google Hands Pentagon Access to AI Models, sparking Firestorm inside tech giant

The deal allows Pentagon access to Google’s AI models for “lawful government purposes,” which includes highly sensitive and potentially classified defense applications. The agreement places Google alongside other major AI players, including OpenAI and xAI, which have also entered similar arrangements with US military institutions.

While contract reportedly includes written safeguards stating the technology is not intended for domestic mass surveillance or fully autonomous weapons without human oversight, critics point out that the language still gives the government wide operational freedom.

Google would not be able to block approved government uses and may be required to assist in adjusting safety settings or system behavior at the Pentagon’s request as per the deal.

As the tech giant defended the arrangement, saying the company continues to oppose the use of AI for unchecked surveillance or autonomous weapons without meaningful human control, the company employees signed an open letter demanding the company refuse to allow its AI systems to be deployed in classified environments. The group argues that such use strips away transparency and oversight, making it impossible for employees to understand how the technology is being used or to intervene if it is misapplied.

Some employees have expressed deep personal concern, warning that classified deployment could enable ethically questionable or harmful applications, including surveillance and lethal military systems developed without accountability to the workforce that built them.

The backlash revived memories of Google’s earlier and highly publicized internal revolt over Project Maven in 2018, a Pentagon initiative that used AI to analyze drone footage. That controversy led to thousands of employee protests and ultimately resulted in Google declining to renew the contract.

For critics inside the company, the core issue is not only military use, but secrecy. Classified deployments, they argue, remove any meaningful visibility into how AI systems are applied in real-world operations, raising fears of unchecked use behind closed doors.

Google shuts down AI Health advice feature ‘What People Suggest’ amid criticism