The U.S. military has come under scrutiny for its use of artificial intelligence in high-stakes operations. Recent reports suggest that AI-generated errors nearly led to an unintended attack on a Chinese ship in the Middle East. According to the accounts, a chatbot used by U.S. intelligence services falsely claimed that a ship was carrying components of a nuclear weapons program. This unverified information was passed along to command structures, prompting plans for an operation to intercept the vessel. The attack was called off at the last minute, narrowly avoiding a possible confrontation with China. The U.S. military has increasingly used AI in various operations, such as analyzing intelligence data, identifying targets for military strikes, and managing logistics. AI systems are capable of quickly processing large volumes of information, which helps speed up decision-making for military leaders. However, this speed can also lead to the spread of incorrect information if the AI-generated data is not thoroughly checked by humans. AI models, known for their tendency to produce "hallucinations"—or false information—can sometimes create misleading conclusions if not properly monitored. A separate incident in February involved a tragic bombing of a school in southern Iran. Two Tomahawk missiles struck the Shajarah Tayyebeh primary school, killing more than 150 people, including at least 123 children. The Pentagon attributed the strike to a series of avoidable errors, including an overreliance on AI and outdated targeting data. This incident underscores the risks of depending too heavily on AI without sufficient oversight or updated information. The AI-related incident involving the Chinese ship reportedly began when an analyst at the Command of Special Operations asked a chatbot to combine open-source data with classified electromagnetic intelligence. The chatbot incorrectly identified the ship’s cargo, and the analyst used the tool again to present the false results in an official-looking summary. This summary was then shared with command structures, prompting plans to intercept the ship. U.S. forces had already deployed aircraft toward the target. Officials decided to re-examine the report the day before the operation and discovered that it had been generated by AI, which had misidentified the ship's cargo. It is unclear whether the AI chatbot involved was a commercial product or a government-modified version of such software. A former senior U.S. official noted that internal tools are often rebranded versions of commercial AI programs. This incident highlights the challenges of using AI in critical military operations, where the benefits of speed and efficiency can lead to overreliance on the technology. Researchers and observers are increasingly concerned about the loss of control as AI systems become more autonomous. Jake Steckler, a researcher at GovAI and former U.S. Army officer, emphasized the importance of recognizing the uncertainty in long-term decision-making models, especially for decisions that could involve the use of force. He argued that rather than removing AI from military decision-making, these incidents should be used to improve AI security. "These tools can be useful in certain contexts with adequate security measures," he said. "But prioritizing the speed of adoption at all costs risks incidents that will only erode military trust in these systems, which, in the end, will only slow their adoption."