Google told staff worried about Pentagon AI deals that the company is 'leaning more' into national security contracts

Google is ramping up its collaboration with the Pentagon, assuring employees that its initiatives with the Department of Defense (DoD) are in line with the company’s AI principles. This shift towards national security contracts was highlighted during a recent town hall meeting at Google DeepMind, where CEO Demis Hassabis and other leaders addressed staff concerns about potential conflicts with the company's ethical guidelines. In response to inquiries about partnerships with the DoD and Boston Dynamics, Tom Lue, Vice President of Global Affairs at Google DeepMind, emphasized the company’s commitment to a rigorous review process. This process evaluates the intended applications of their technology and assesses safeguards for safety, responsibility, and privacy. Notably, Lue noted that Google had revised its AI principles in 2025, removing a prior commitment to refrain from developing militarized technology or engaging in surveillance. Lue stated, "The north star for the analysis is whether the benefits substantially exceed the risks." This statement comes amidst a growing debate in the tech industry regarding the involvement of companies in military and surveillance activities. Recently, Anthropic, another AI company, found itself in conflict with the DoD after setting strict boundaries on how its technology could be utilized, leading to a lawsuit against the Pentagon over claims of being blacklisted. As tech giants like Google, Amazon, and Oracle increasingly seek defense contracts, Lue indicated during the January town hall that employees should anticipate more engagements with government entities regarding national security matters. He mentioned ongoing discussions focused on cybersecurity and biosecurity risks, reinforcing the company's direction towards involvement in such sectors. In a follow-up to these engagements, a Google DeepMind representative directed attention to a recent blog post detailing the company’s latest Pentagon contract, which involves AI applications for administrative tasks like document drafting and project planning. Hassabis expressed his confidence in the balance Google is achieving in its defense partnerships, stating, "It’s important for us to work with democratically elected governments to leverage our capabilities for global safety." After a hiatus from military contracts in 2018 due to employee protests, Google re-established its relationship with the Pentagon last year, securing contracts for AI and cloud services, culminating in a recent agreement to implement AI agents in unclassified defense networks. Lue clarified that these contracts are primarily aimed at clerical tasks, not combat operations. Despite the reassurances, there remains trepidation among some employees regarding the potential misuse of technology, particularly concerning contracts with the Israeli government. Recently, an open letter signed by employees from Google and OpenAI urged their companies to prevent their technologies from facilitating autonomous weapons or mass surveillance. A whistleblower complaint also raised concerns about Google allegedly breaching its own policies by aiding an Israeli drone contractor with analysis tasks. Google responded that the scope of this assistance was negligible. Furthermore, Google DeepMind has been collaborating with Boston Dynamics to integrate its AI systems into robotic platforms, with assurances that these technologies will not be used for weaponization. Lue reiterated the company’s extensive review protocols during the town hall, aiming to maintain ethical standards in its technological advancements.