AI Nearly Starts a War, and Weapons Reach Orbit
How this was made Verified AI
Every Intellegix briefing is generated from that day's broadcast and run through automated checks before it publishes — with a human paged on any flag. Here is the trail for this edition.
CNN reported this week that an AI-generated false report nearly triggered a direct military clash between the United States and China. Details remain classified, but the broad outline is that an AI system produced an intelligence assessment — assessed as credible at some point in the decision chain — describing a Chinese military action that had not occurred. Forces were moved and decision trees were activated before the report was determined to be AI-fabricated and the situation was walked back. It is the kind of incident that will be discussed in national security circles for a long time, even as it receives less attention than domestic political news.
The implications extend well beyond the immediate incident. Military decision-making under time pressure is already among the most error-prone human activities studied. Introducing AI systems capable of generating plausible, internally consistent, detailed intelligence products describing events that never happened adds a new category of failure mode that existing military verification procedures were not designed to handle. Those procedures were built on the assumption that fabricated intelligence is difficult to produce. Generative AI makes it trivial.
The same underlying technology appears in the Kremlin's deepfake campaign targeting November's midterms, applied to a different domain. The generative models capable of producing a convincing celebrity endorsement video are the same models capable of producing a convincing intelligence product describing a military event that did not occur. The defense mechanisms are similarly immature in both domains.
The week's most quietly consequential military disclosure came from the Air Force Secretary, who confirmed that Space Force has deployed weapons to orbit — the first public acknowledgment that the United States has operational weapons in space, not merely reconnaissance satellites or communications assets. The Outer Space Treaty of 1967 prohibits weapons of mass destruction in orbit but does not prohibit conventional weapons. The disclosure, if accurate, represents a significant escalation in the militarization of space that will prompt responses from China and Russia, both of which maintain active counter-space programs.
Trump's AI Force announcement, comparing it to Space Force, takes on a different character in that context. Space Force began as a widely mocked proposal and became a real military branch with a genuine operational mission — including, apparently, orbital weapons deployment. An AI Force would presumably be tasked with both offensive and defensive AI capabilities in military contexts. Given the near-miss described above, the question of what that force's rules of engagement look like is not abstract. The Trump administration simultaneously hosted a UN event on AI governance, a multilateral positioning exercise that other governments can evaluate against the domestic reality: a blocked federal AI regulator, an AI Force with unclear authorities, and a Senate floor where a kill switch bill was just defeated.