The recent disagreement between Anthropic and the US Department of Defense regarding the boundaries of AI's military use has drawn significant attention.
Anthropic firmly adhered to two red lines: never integrating AI into "fully automated lethal weapon systems" without substantial human intervention, and never using AI to monitor Americans. However, the Pentagon demanded completely unrestricted military use rights, and after being refused, it terminated the contract and labeled the American company a "supply chain security risk."
Nevertheless, Anthropic's uncompromising stance propelled Claude to the top of the App Store charts in over twenty countries worldwide.
In fact, in April 2024, Israeli independent journalist Yuval Abraham published an in-depth report exposing the "Lavender" AI system used by Israel during its operations in Gaza. This system automatically generated assassination lists by capturing massive amounts of surveillance data, identifying as many as 37,000 potential targets in the first few weeks of the war. The article states that the military was aware of the AI's 10% false positive rate, yet intelligence officials spent an average of only 20 seconds reviewing each target, effectively reducing humans to mere tools for blindly stamping approvals by machines. One intelligence official even coldly admitted that they preferred to trust statistical machines rather than emotionally driven soldiers.
In this process, ethical standards steadily declined. To maximize efficiency, the military allowed the killing of 15 to 20 innocent civilians to eliminate a low-ranking Hamas member identified by AI; this "civilian death quota" was even extended to hundreds if the target was a high-ranking commander. Often, this was done at night when the target returned home, combined with an automated tracking system called "Where Did Dad Go?", to save costs by dropping unguided "duds," leveling the entire apartment building along with the target's family.
AI is hot, and we discuss its useful changes in our daily lives—generating music and images from videos, turning ideas into code… But beyond these, in vast areas, AI is also triggering extremely drastic and even brutal changes. Of course, we are powerless to change anything when faced with state apparatus, military, or war. But at least we can know what is still happening in this world.
