Tech workers should distinguish between the role of technologies like AI and AR on the battlefield and general questions of whether US war-making is right
Lucas Kunce / New York Times : Tweets: @paul_scharre , @athertonkd , @ebkania , @dellsystem , @nytopinion , @matthewstoller , and @matthewstoller Tweets: Paul Scharre / @paul_scharre : Great article by a Marine on the responsibility US tech workers have for working constructively to ensure responsible military use of AI. “Dear Tech Workers, U.S. Service Members Need Your Help” https://www.nytimes.com/... Kelsey D. Atherton / @athertonkd : This is a trap: https://www.nytimes.com/... “For me, it's hard to understand why tech employees would not want to help their fellow Americans survive on the battlefield and accomplish their missions in the safest and least damaging way possible.” Try to understand harder. Elsa B. Kania / @ebkania : “Tech companies and their employees have immense power and knowledge in this country. Using those assets to create a real discussion about when America uses force would be a much more enduring legacy.” https://www.nytimes.com/... @dellsystem : i think the conclusion here is that the government is very capable of manufacturing support for endless war, rather than that our democratic institutions are *working* pic.twitter.com/v4X6ZfZD9G @nytopinion : When tech employees protest against working with the military, it “has the unintended effect of imperiling not only the lives of service members, but also the lives of innocent civilians whom I believe these workers want to protect,” writes Lucas Kunce https://www.nytimes.com/... Matt Stoller / @matthewstoller : “Tech workers might not realize that their opposition to the work their companies do on military technology does not change the decision-making of the American leaders who choose to go to war, and therefore is unlikely to prevent any harm caused by war.” https://www.nytimes.com/... Matt Stoller / @matthewstoller : Google employees should consider these points when thinking about working with the Pentagon. Warfare tech and AI isn't necessarily about killing, and weakening the military institutions isn't the right way to change *war policy.* https://www.nytimes.com/...
Context & Ripple Effects
Lucas Kunce's op-ed lands mid-fight over whether tech workers should touch military AI at all. The resistance has real history: employee outcry ended Google's work on Project Maven, and the related coverage shows the industry split — Clarifai's CEO arguing small startups need no internal ethics officers and would supply autonomous weapons, while Washington weighs how much lethal decision-making to delegate to machines at all.
Kunce's move is to reframe the question: separate the battlefield role of tools like AI and AR from judgments about US war-making itself. Kelsey Atherton calls that framing a trap, and Matt Stoller argues worker opposition rarely changes decisions to go to war — while Elsa Kania urges employees to use their power to force debate over when force is used. The stakes became concrete when Google returned to pursue a major Pentagon cloud and AI contract three years after the Maven walkout.
First-order effects
- Tech workers at defense-adjacent projects face a sharpened choice: blanket refusal, or engagement aimed at responsible use — with Kunce claiming service members' survival depends on that expertise and Atherton warning the framing co-opts dissent.
Second-order effects
- Google's renewed Pentagon bid suggests employee walkouts raise costs but do not settle corporate strategy, pushing the contest inside companies — over ethics review, transparency, and who speaks for the workforce — rather than over whether to bid at all.
Third-order effects
- If the tool-versus-war distinction holds, military AI governance shifts from veto politics to participation: workers, ethicists, and regulators contesting how systems are built and constrained, as the Pentagon's own autonomy ethics questions move from abstract to operational.
The trend: Silicon Valley's relationship with the Pentagon is shifting from employee-veto standoffs toward negotiated participation, with the fight moving to how military AI is governed rather than whether it gets built.