How AI ‘almost started a war’ between US and China with false intel: Report

Earlier this year, US military forces were set to intercept a Chinese ship in West Asia that was allegedly carrying components for a nuclear weapons program, according to an intelligence report. The US military prepared to board the vessel and dispatched military planes to the scene, all based on an intelligence report generated using artificial intelligence (AI).

However, it was later revealed that the report was entirely false, circulating due to an AI-generated error. The report was prepared by a special operations command analyst who used a chatbot to analyze the ship's manifest. The chatbot inaccurately identified the materials on the ship, leading to false intel that nearly escalated into a military conflict between the US and China.

The AI used to generate the report was a combination of open-source intelligence and secret signals intelligence held by the US government. The incident highlights the risks of relying on AI for critical military decisions, especially in times of war. Different parts of the US government use various tools and safety standards, with no unified standards for verifying AI-generated information.