Google has quietly abandoned its long-held promise not to use artificial intelligence for weaponry, declaring that “free countries” should be able to harness the technology for national security purposes.
The policy shift emerged when two senior figures, James Manyika (Google-Alphabet’s senior vice-president) and Sir Demis Hassabis (chief executive of Google DeepMind), co-authored a blog post confirming the technology giant’s commitment to “support national security” in an increasingly tense geopolitical race for AI leadership.
The new stance marks a reversal of Google’s 2018 assurance that it would never use AI tools “whose principal purpose or implementation is to cause or directly facilitate injury to people”. That vow was adopted after staff walked out in protest against a Pentagon drone project. In the wake of the latest announcement, critics argue that Google risks undermining its values, noting that the company also removed its “don’t be evil” motto from its code of conduct when it restructured under parent entity Alphabet in 2015.
Manyika and Hassabis justified the change by highlighting the accelerating pace of AI innovation and pointing to the threat posed by China’s growing military interest in the technology. Beijing has earmarked AI as the future “revolution in military affairs” and is reportedly using the technology to develop advanced autonomous weapons systems. DeepSeek, a Chinese-developed AI chatbot, has already achieved results that in some tests surpass western competitors, fuelling concerns of a “Sputnik moment” in the sector.
Since the onset of generative AI, Google has faced internal strife over its links to the defence sector. In 2018, staff petitioned executives to withdraw from a US military drone initiative. There has also been friction over partnerships with foreign governments, notably in Israel. Adding to the debate, Geoffrey Hinton – dubbed the “godfather of AI” – quit Google in 2023, warning the technology could one day threaten humanity itself.
While critics lament Google’s apparent climbdown, company chiefs maintain that democracies must lead in AI development. They argue that collaboration between companies, governments and organisations that “share these values” will help ensure AI is harnessed responsibly – including for national defence.
