Google has updated its AI safety policy, removing its previous commitment not to use AI for military and intelligence purposes. The original guidelines, introduced in 2018, were revised and published on the company’s official blog, as reported by The Washington Post.
Previously, Google had pledged to avoid four categories of AI use: weapons, surveillance, technologies “capable of causing harm,” and projects violating international law and human rights. These restrictions set Google apart from other AI market leaders, such as Microsoft and Amazon, which have long collaborated with the Pentagon. OpenAI and Anthropic have also partnered with U.S. defense contractors like Anduril and Palantir.
AI’s Growing Role in Defense
Experts believe Google’s decision highlights AI’s increasing significance in U.S. national defense. Michael Horowitz, a political science professor at the University of Pennsylvania, described the shift as a logical step, given AI’s expanding role in military applications. “Google’s announcement is further evidence that the relationship between the U.S. tech sector and the Department of Defense continues to grow closer, including with leading AI companies,” Horowitz stated.
However, some critics remain skeptical. Lilly Irani, a former Google employee and professor at the University of California, San Diego, argues that “the company’s promises to uphold international law and human rights are often empty words.”
A Shift in Priorities
Google originally introduced its AI ethics guidelines following employee protests against Project Maven, a Pentagon initiative that used AI to analyze drone footage. Thousands of employees signed a petition demanding the company withdraw from military projects. In response, Google not only refused the contract but also committed to avoiding weapons development.
Yet, the recent policy change indicates a shift in priorities, notes NIX Solutions. While the full implications remain unclear, we’ll keep you updated on further developments regarding Google’s evolving AI policies.