The US military nearly launched an operation against a Chinese-flagged vessel in the Middle East after receiving flawed intelligence from an artificial intelligence tool. The incident occurred during the ongoing conflict with Iran and involved preparations to intercept the ship, including moving armed personnel into position. Military aircraft were already in the air and armed service members were preparing to board the vessel, according to CNN.
Military officials discovered the issue shortly before the planned operation was set to begin when they reviewed the intelligence more carefully. The faulty information had been generated by a Special Operations Command analyst who used an AI chatbot in their analysis. The AI tool incorrectly identified the cargo aboard the ship, claiming it was carrying components for a nuclear weapons program.
One source familiar with the episode described the intelligence report to CNN as “entirely false.” The source said the incident almost started a war because an attempt by US troops to board a Chinese vessel could have triggered a confrontation between the two countries.
The intelligence originated from US Special Operations Command Pacific in Hawaii and combined open-source data with classified signals intelligence. An analyst queried the chatbot about intelligence related to the ship’s manifest. The AI system then combined public information with secret signals intelligence held by the government before reaching the incorrect finding. The analyst used AI again to turn the information into a standard intelligence report that was circulated through military channels.
CNN reported that it could not determine whether the chatbot was a commercial product or a tool developed for the US government. US Special Operations Command Pacific and the Pentagon did not respond to CNN’s requests for comment on the incident.
This event highlights the increasing use of AI within military and intelligence operations and raises serious concerns about relying on automated systems during wartime. CNN reported that such incidents could lead to dangerous decisions based on misleading outputs from AI and other automated tools. Sources told CNN that there is not a consistent military-wide system governing how all AI tools are used and checked.
The situation was not limited to this single case, as the Pentagon continues to rely heavily on AI systems like Palantir’s Maven Smart System for targeting. The Pentagon designated Maven as a core military program this year. The system processes information from satellites, drones, sensors and intelligence sources and can help identify potential battlefield targets. Human operators remain responsible for lethal decisions.
The Department of Defense has promoted AI use under Defense Secretary Pete Hegseth, despite growing concerns about the risks involved. Hegseth has said he wants the military to become an “AI-first war fighting force” and has called for faster experimentation and fewer barriers to adopting the technology.
A separate incident on February 28, 2026, saw a US Tomahawk missile strike an elementary school in Minab, Iran. A United Nations fact-finding mission later said it had reasonable grounds to believe the United States was responsible for the attack. The UN said 156 civilians were killed, including 120 children. Earlier Iranian casualty reports had placed the death toll higher.
The Minab strike was linked to outdated intelligence rather than an AI hallucination. CNN later reported that senior US military commanders approved some strikes even though military databases contained warnings that intelligence on potential Iranian targets was severely outdated. Three sources told CNN that commanders ignored some of those warnings as officials rushed to prepare targets at the start of the conflict.
Human Rights Watch reported that US Central Command officers created the coordinates for the strike using outdated information provided by the Defense Intelligence Agency. The available evidence does not establish that an AI system invented the target. Questions remain about how AI-assisted targeting systems handled or displayed the stale information.
The military’s Central Command has been using AI tools to process massive amounts of intelligence data. Maven helped US forces process targeting information during the Iran campaign. CSIS reported that Maven assisted US operations against more than 1,000 targets during the first 24 hours of the conflict.
These systems can produce targeting recommendations quickly, but they are not immune to errors when fed incorrect or outdated information. The Minab strike showed that even accurate software cannot correct intelligence that has not been properly updated or verified.
Iran-linked actors have reportedly used similar AI technology to develop targeting strategies against US naval assets, showing that adversaries are also adopting the technology. Anthropic reported that an Iran-linked threat actor used Claude to collect open-source information about US Navy forces and develop targeting recommendations.
Experts now question whether the current framework for AI oversight is sufficient to prevent future errors like those seen in both cases. US and Chinese security experts have proposed safeguards that include maintaining meaningful human control over sensitive military uses of AI and establishing channels for the two countries to communicate during AI-related crises. The proposals have not been formally adopted by either government.
Others argue that operators must treat AI results as information that still requires independent verification. CNN reported that military officials have raised concerns about inconsistent rules for checking AI-generated intelligence before it moves through the chain of command.
The Pentagon now faces growing scrutiny over its approach to integrating AI into operations, especially when human judgment is not properly enforced. Military officials and outside researchers have stressed that human oversight remains important because AI systems can produce answers that appear convincing even when they are wrong.
In the meantime, a US military operation was narrowly prevented from taking place due to an AI-generated error that could have led to serious consequences, according to a report from Gizmodo. The event underscores how quickly automated systems can mislead decision-makers and emphasizes the need for better safeguards in high-stakes environments, Israel National News reported.
IMAGE: EJ Hersom, DoD CC0