Google has just signed a confidential agreement with the American government to use its artificial intelligence in the context of classified defense affairs. Over 600 employees, including vice presidents, recently signed a letter asking for the negotiations to be stopped, expressing their concerns about the ethical implications.
After the highly criticized Maven project, Google had promised to never collaborate with the American government for military purposes again. In February 2025, the American company changed its policies. Today, it has signed a confidential agreement with the American government for the use of its AI in classified defense affairs. More than 600 employees, including vice presidents and engineers from Google DeepMind, have opposed this collaboration.
A letter to demand the cessation of negotiations
At the beginning of April, The Information media revealed that Google was in talks with the Pentagon for the use of its artificial intelligence in classified defense affairs. On April 27, over 600 Google employees addressed a letter to Alphabet CEO, Sundar Pichai, asking the company to refuse the use of artificial intelligence for classified military affairs. The signatories include employees, as well as directors, vice presidents, and engineers from Google DeepMind. This letter carries significant weight.
In this letter, an anonymous employee explains: “Currently, there is no way to ensure that our tools will not be used to cause terrible harm or undermine individual freedoms, away from view.” That is precisely the problem. Google has requested that its AI not be used for mass surveillance in the United States or in deadly attacks, but how can we know if these rules will be applied, especially when it comes to classified defense affairs?
Google changes course and signs a confidential agreement with the American government
Google already works for the American government, but only in public affairs. In 2017, the American company had collaborated with the Pentagon for the Maven program, which used artificial intelligence to analyze images captured by military drones automatically. At the time, over 3,100 employees had signed a petition to prevent this military collaboration. Google had promised to never repeat this type of collaboration.
However, things have changed. On February 4, 2025, Google updated its general principles regarding artificial intelligence, removing its commitment not to apply this technology for military purposes. In April 2026, Google has signed a confidential agreement with the Pentagon for the use of its artificial intelligence in classified defense affairs, according to The Information media. The agreement would authorize the Pentagon to use Google’s AI for “any legal government purpose.”
The American government seeks to diversify its AI
This agreement comes at a specific time when the American government is actively seeking to diversify its artificial intelligence. The startup Anthropic, which refuses to use its AI for military or surveillance purposes, is in conflict with the American government. The CIA’s deputy director explained that the American intelligence service “cannot allow the whims of a single company [Anthropic]” to restrict its use of AI. The American army has turned to Sam Altman (ChatGPT), who is less stringent on these issues, and now turns to Google, which seems to have decisively changed its policies in this regard.
Source: The Information
