US military AI made a mistake in identifying a Chinese ship

The US military intelligence system, using artificial intelligence algorithms, generated a false report of the presence of nuclear components on board a Chinese ship. The error was identified by officers just before making a decision to start retaliatory measures, which prevented a potential international conflict.

According to the report, the AI tool analyzed satellite images and radio intercept data, mistakenly classifying civilian cargo as prohibited military materials. The algorithm made a hallucination, interpreting random interference on the image as specific containers for transporting radioactive elements. During the subsequent investigation, it was found that the system did not have a sufficient level of verification of data from open sources.

The incident highlights the critical risks of implementing automated decision-making systems in the defense sector. Experts note that the lack of "human control" mechanisms at the stage of primary data processing led to the fact that false information was transmitted to the headquarters as a confirmed fact. In discussions of the incident, specialists point to the need to implement strict verification protocols for all AI agents working with intelligence data.

The US Department of Defense has initiated a review of data processing algorithms to exclude similar failures in the future. In the current situation, military agencies are forced to re-examine their trust in neural network models, which tend to generate convincing but actually incorrect conclusions when analyzing complex geopolitical situations.