Due to AI error, US almost launched military operation against Chinese vessel
A report prepared with the help of AI claimed that the ship was carrying components for nuclear weapons.
In spring, during the war with Iran, an intelligence report spread through the US army: a Chinese ship in the Middle East region was allegedly carrying components for nuclear weapons. The military began preparing to intercept the vessel. Planes were in the air, and armed teams were preparing for boarding.
Just before the operation, it turned out that the report was compiled by an analyst from the US Special Operations Command in the Pacific using artificial intelligence, and the chatbot mistakenly identified the cargo, CNN reports. What specific cargo was mistakenly identified is unknown.
The operation against the Chinese ship risked escalating into a conflict between the two states. It is unknown whether it was a commercial AI chatbot or a government product. The bot combined open data with classified signal intelligence, and the analyst then formatted the conclusions into a standard intelligence report, again with the help of AI. The error was noticed by more experienced colleagues who double-checked the data.
The Pentagon is now seeking to implement AI almost everywhere: from intelligence analysis and target selection to budgeting and logistics. In January, Secretary of War Pete Hegseth unveiled the "AI Acceleration Strategy" – the goal is to keep pace with China and other adversaries.
According to one CNN source, similar "hallucinations" have happened in intelligence before.
Analysts, especially younger ones, are accustomed to these tools and are more inclined to trust them uncritically. The pressure to do everything faster increases the risk of errors.
Comments