Google employees urge CEO to stay out of secret Pentagon AI projects

Google is facing a new internal conflict over military contracts. About 600 employees have demanded that the CEO drop out of secret Pentagon projects if they could use AI for surveillance and autonomous weapons.

A letter to management

Several hundred Google workers sent a letter to the head of the company asking it not to draw the business into closed US defense programs. Their main objection concerns scenarios in which Google's technology could be used for mass surveillance or to build systems capable of making lethal decisions without direct human involvement. For part of the team, this is not merely a question of reputation but a line the company should not cross, even for large contracts and strategic ties with the government.

The authors recall that a similar dispute already played out inside Google in 2018. Employee protest against the company's participation in a military project became so visible that management promised not to develop AI that violates the company's own principles. The current appeal shows that trust in those promises has weakened: employees believe management's position has changed, and the earlier restrictions no longer look firm or unambiguous. In effect, they are demanding that the old commitments be reaffirmed in a more specific and verifiable form.

Why the dispute is back

The new conflict did not come out of nowhere. In recent years, generative AI and computer vision systems have become far more powerful, and military interest in such technology has risen sharply. This means that even general-purpose cloud services, image analysis models and data processing tools can be used for tasks that once seemed remote from civilian business. That is why Google's workers are now talking not about a hypothetical risk but about quite real use cases.

Employees are worried not only by general talk of cooperation with the state but by specific classes of tasks and ways of applying AI. They fear that the company's internal rules may be interpreted too broadly, and that decisions on sensitive contracts will be made without a clear explanation to the team, without external reporting and without an independent assessment of the consequences for society and for the company itself. The letter highlights several key risks:

• use of Google's AI in mass surveillance systems
• use of models in secret military projects without transparent reporting
• the company's participation in developing autonomous combat systems
• erosion of the principles Google publicly adopted after the 2018 crisis

The moment is awkward for Google also because the whole AI market is rapidly moving closer to the public sector. Big tech companies sell clouds, models and infrastructure to businesses, government agencies and defense customers at the same time. Against this backdrop, drawing a clear line between a "neutral platform" and direct involvement in sensitive programs is getting harder, and the internal ethics debate is no longer abstract. This dispute therefore looks less like a local conflict than a symptom of the whole industry.

What has changed since 2018

In 2018, the internal employee protest became one of the loudest examples of how a tech company's staff can influence strategic decisions from below. Google then not only faced a wave of discontent but was forced to explain publicly where the limit lies for its AI work. Afterward, the company set out AI principles meant to ease concern inside the team and beyond.

The situation now looks different. The AI market has become much more competitive, and government and defense contracts matter more to providers of cloud and computing infrastructure. As a result, employees take any revision of the earlier limits harder: it signals that ethical frameworks may give way under pressure from commercial interests, geopolitics and the race for technological leadership. If the company does not explain the boundaries anew, each new contract of this kind will set off another round of distrust.

What this means

The letter from 600 employees shows that the main dispute over AI is no longer only between companies and regulators, but also inside the corporations themselves. For the market, it is a signal: the deeper AI platforms move into defense and government work, the sharper the conflict will be between business growth, transparency and the limits of acceptable use of the technology.

For Google, it is also a test of how far its public principles actually constrain the company's real strategy.