The United States "almost started a war" against China after an AI chatbot wrongly reported a Chinese ship was carrying nuclear weapons components, according to a new report.
The incident happened this spring, when the vessel was spotted in the Middle East in the midst of America's war with Iran. The US was preparing to board the vessel and planes were circling, with plans to intercept the ship underway.
But, once officials had a more detailed look into the intelligence report that led to the large-scale response, they realised it had been generated with the help of AI, CNN reports.
Analyst's AI error
An analyst is said to have used a chatbot that wrongly identified the nature of the Chinese ship's cargo, suggesting the vessel was transporting components of a nuclear weapons programme.
What exactly the vessel was transporting remains unclear to this day, with one source telling CNN that the intelligence report was "entirely false" but "almost started a war" between the US and China.
AI systems under scrutiny
It is also unclear whether the AI chatbot used by the US analyst was one that is commercially available or one that has been developed by the US government.
However, a former senior US official who has knowledge of the AI systems used by the country's military said the internal tools are "mostly just copies of the commercial stuff wearing lipstick".
