A Chatbot-Built Intelligence Report Nearly Sent US Troops Onto a Chinese Ship
An AI-assisted report falsely tied a Chinese ship's cargo to a nuclear program during the Iran war. Senators now want inspectors general to examine how the Pentagon checks machine-made intelligence.
An AI-assisted report falsely tied a Chinese ship's cargo to a nuclear program during the Iran war. Senators now want inspectors general to examine how the Pentagon checks machine-made intelligence.
US forces prepared to board a Chinese vessel in the Middle East this spring after an AI-assisted intelligence report wrongly identified its cargo as nuclear weapons components, CNN reported, citing four people familiar with the episode. Military aircraft were airborne and an armed boarding team was ready when officials found the flaw.
Frequently Asked Questions
What happened with the Chinese ship?
This spring, during the war with Iran, an AI-assisted US intelligence report wrongly said a Chinese ship in the Middle East carried nuclear weapons components. US forces prepared to board it before officials caught the error, CNN reported.
Which AI tool produced the error?
CNN could not determine whether the chatbot was a commercial product or a government tool. There is no indication in its reporting that ChatGPT Mil or Grok for Government, added to GenAI.mil on Aug. 31, was involved.
How far did the operation get?
Military planes were in the air and armed personnel were preparing to board, according to CNN's sources. Officials examined the report just before the planned operation and found it was AI-generated and false.
What are lawmakers asking for?
Senators Mark Warner, Jack Reed and Chris Coons asked Defense Secretary Pete Hegseth and DNI Jay Clayton on Sept. 19 for an immediate inspector general investigation into AI-related targeting errors, including the February strike in Minab, Iran.
Does the Pentagon have a standard for verifying AI-generated intelligence?
No single standard exists, US officials told CNN. Different parts of the military and intelligence community use different tools under different rules and safety standards.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Read next — from the Lexicon
Below this point
4 more sections.
