TL;DR
- Interception Plans: US forces prepared to intercept a Chinese ship over a false AI-assisted intelligence report, according to CNN.
- Formal Report: An analyst used AI to assess the cargo and again to package the findings for military recipients.
- Human Intervention: Experienced analysts checked the underlying information and stopped the operation, CNN reporter Katie Bo Lillis said.
- Unknown Tool: The chatbot’s identity and any disciplinary consequences remain unknown.
US forces prepared to intercept a Chinese ship during the Iran war after an AI-assisted intelligence report wrongly identified its cargo as components of a nuclear weapons program, CNN reported on September 18. Armed personnel were preparing to board and military aircraft were already airborne before officials caught the error, according to the network’s sources.
The episode occurred in spring 2026. CNN cited four sources familiar with the interception plans; two described boarding preparations, while one of those sources and another described the aircraft.
From Cargo Assessment to Possible Military Intervention
A special operations command analyst asked a chatbot about intelligence concerning the ship’s manifest, its cargo record. That intelligence originated with US Special Operations Command Pacific in Hawaii. CNN said the bot combined publicly available information with classified intelligence held by the government and incorrectly identified the cargo.
The analyst then used AI to turn the findings into a standard intelligence report and circulated it. CNN described that format as one military officials trust.
In a CNN broadcast later that day, reporter Katie Bo Lillis described the intervention more fully. Experienced analysts with expertise in the subject examined the information behind the report just before the planned operation. They recognized that the cargo identification was wrong, and the operation was called off.
Lillis characterized the incorrect identification as a hallucination, the term for an AI system generating false information.
Whether anyone checked the assessment before it circulated, and how the later expert review began, remain unknown.
Pentagon Declines to Discuss Specifics
CNN could not establish what the cargo actually was or whether the chatbot was a commercial product or an internal government system. The exact prompt and any disciplinary consequences for the analyst remain unknown.
The Pentagon and US Special Operations Command Pacific did not answer CNN’s requests for comment. A Defense Department official separately told the Daily Beast that operational security and policy prevented discussion of specific intelligence reports, alleged operational planning or ongoing assessments.
The official also said the department remained committed to rigorous verification of intelligence inputs and responsible integration of new technologies.
Faster Decisions Need Reliable Intelligence
Defense Secretary Pete Hegseth’s January push to expand military AI use preceded the incident. The stated rationale was to make decisions faster and avoid falling behind adversaries using similar technology. In his January remarks on the Pentagon’s technology agenda, Hegseth demanded models should be factually accurate and relevant to military missions.
The Defense Intelligence Agency, which supplies intelligence to military planners and policymakers, asked industry about tools to test and validate AI. Its request sought methods for checking systems throughout development, deployment and operational monitoring, including tools supporting analysis and targeting.
The agency was particularly interested in automating those checks throughout a system’s working life.