A false AI assessment reportedly pushed the U.S. military toward boarding a Chinese ship, before a late review exposed the cargo finding as wrong.

A U.S. special operations analyst used an AI chatbot this spring to examine the ship’s manifest, the Daily Mail reported. The system wrongly identified the cargo as parts for a nuclear-weapons program.

The analyst then used AI to turn that finding into an intelligence report. It circulated across the military and prompted planners to prepare an interception and boarding operation in Middle Eastern waters.

Aircraft were already in the air while soldiers prepared for the mission. Officials reviewed the assessment before the operation began and discovered that AI had helped produce it.

That review found the technology had misidentified the material aboard the vessel. An unnamed source described the intelligence report as “entirely false” and said the episode “almost started a war.”

Human review caught the AI mistake late in the process

The military did not board the ship, based on the information available. But the error traveled far enough to shape an operational plan and put personnel in motion before reviewers stopped it.

The sequence exposes the central risk in military AI. A computer can produce an answer quickly, but speed offers no protection when the answer itself is wrong.

The Pentagon has used AI to examine raw intelligence, identify possible strike targets, manage logistics, and help balance budgets. Those tools now touch decisions with consequences far beyond an office spreadsheet.

That makes accountability more pressing amid the broader debate over Hegseth’s Pentagon leadership. Innovation can help the armed forces, but commanders still need reliable checks before machine-created claims drive real-world action.

An unnamed source warned that AI use in targeting was increasing without clear rules for how human oversight would prevent civilian deaths or accidental attacks on friendly forces.

“AI in targeting is definitely something that is ramping up and there is no real guidance for how having a human in the loop will prevent civilian casualties or fratricide,”

That warning goes beyond this Chinese vessel. If analysts and planners cannot trace how a system reached its answer, human review risks becoming a final-minute safeguard instead of a firm requirement.

Hegseth pushed faster military AI adoption in January

Secretary of War Pete Hegseth announced an initiative in January to speed AI adoption across the armed forces. His stated goal was to cut barriers and keep the United States ahead in military uses of the technology.

Hegseth said during that announcement:

“We will unleash experimentation, eliminate bureaucratic barriers, focus our investments and demonstrate the execution approach needed to ensure we lead in military AI,”

The Chinese ship incident does not establish that Hegseth’s initiative caused the analyst’s error. It does show why faster adoption must come with firm standards for sourcing, review, and final approval.

That operational question carries more weight than the political disputes surrounding the secretary, including a shelved impeachment push. Congress and military leaders should focus on rules that keep untested output from moving unchecked through the chain of command.

An unnamed former U.S. official offered a blunt assessment of the government’s internal AI systems.

“The internal tools are mostly just copies of the commercial stuff wearing lipstick,”

The source did not identify which chatbot the analyst used. It remains unclear whether the system was a commercial product or an internal Pentagon tool.

Key questions remain unanswered after the aborted mission

No exact date was provided for the manifest review, the intelligence report, or the planned interception. The ship’s name, its actual cargo, and the military units involved also were not disclosed.

The available details do not identify which officials caught the error. They also do not explain what review rules applied before aircraft took off and soldiers began preparing.

Those gaps matter as Hegseth manages other personnel and command disputes, including his decision to pull four promotion candidates. Clear responsibility matters most when a flawed assessment advances toward action.

The revelations emerged less than a week before Chinese President Xi Jinping was scheduled to visit Washington and meet President Donald Trump. Their planned discussions included AI, trade, critical minerals, and other issues.

The timing raised the diplomatic stakes, but the basic failure was operational. An AI-created cargo claim entered military channels, spread across the force, and reached the mission-planning stage before officials rejected it.

Leadership changes will not settle that problem by themselves, even as attention turns toward possible choices for the next Army secretary. The military needs a review system that tests AI claims before personnel and aircraft begin moving.

America should lead in military technology without surrendering judgment to it. Machines may assist the mission, but accountable people must remain in command.