Google Steps Up Where Anthropic Stepped Back
In a move that's reigniting the debate over AI ethics in the defence sector, Google has signed a new contract with the U.S. Department of Defense, expanding the Pentagon's access to its artificial intelligence tools. The deal comes after Anthropic — the AI safety company behind the Claude models — reportedly refused to allow the DoD to use its AI for domestic mass surveillance and autonomous weapons programs.
The contrast is striking: two of the biggest names in AI, taking dramatically different stances on one of the thorniest questions in the industry.
What Anthropic Refused
According to reporting from TechCrunch, Anthropic drew a hard line when approached by the Department of Defense, declining to grant access to its AI systems for use cases involving domestic mass surveillance or autonomous weapons. The company has long positioned itself as a safety-first AI lab, and this refusal appears to be a direct extension of that philosophy into its business decisions.
It's a notable stance at a time when AI companies are under enormous pressure — financial and political — to capture lucrative government contracts. Defence contracts can be worth billions, and the Pentagon has been aggressively expanding its AI capabilities across virtually every branch of the military.
Google Fills the Gap
Google, by contrast, has moved in the opposite direction. The tech giant has signed an expanded agreement with the DoD, giving the department broader access to its AI tools and infrastructure. The specifics of what capabilities are included haven't been fully disclosed, but the timing — following Anthropic's refusal — is hard to ignore.
This isn't the first time Google has navigated controversy over military AI work. In 2018, the company faced intense internal backlash — including employee resignations — over its involvement in Project Maven, a Pentagon initiative that used AI to analyse drone footage. Google eventually declined to renew that contract. The new deal suggests the company has recalibrated its appetite for defence partnerships in the years since.
A Defining Moment for AI Ethics
The divergence between Google and Anthropic puts a spotlight on a question that the AI industry hasn't yet answered cleanly: what should the limits of AI be when it comes to military and surveillance applications?
For Anthropic, the answer involves explicit red lines around domestic surveillance and autonomous weapons — systems capable of making lethal decisions without human oversight. For Google, the calculus appears to weigh commercial opportunity differently, though the company would likely argue its tools are used for logistics, data analysis, and other non-lethal applications.
Critics of the Google deal argue that once AI tools are inside a defence apparatus, controlling how they're used becomes extraordinarily difficult. Supporters counter that American AI capabilities in defence are inevitable, and it's better to have safety-conscious companies involved than to cede the space entirely.
What It Means Going Forward
The episode underscores a growing fault line in the AI industry between companies that see government and military contracts as core business, and those that are trying to build their identities around ethical constraints — even when it costs them revenue.
As AI becomes more capable and more embedded in critical systems, these decisions will only become more consequential. Whether Anthropic's refusal sets a precedent others follow, or whether Google's approach becomes the industry norm, remains to be seen.
Source: TechCrunch — Google expands Pentagon's access to its AI after Anthropic's refusal
