Subscriptions
Read our Monthly Magazine
And support our mission to provide fearless stories about and outside the media system
The US military “almost started a war” with China this spring due to a faulty AI-generated intelligence report that falsely concluded that China was transporting nuclear materials aboard a ship in the Middle East, according to an alarming new report.
Citing four unnamed sources, CNN reported on Friday that the US military was so confident in the information contained in the AI-generated intelligence report that it made plans to intercept the ship and even started carrying out those plans. Armed members of the military were reportedly getting ready to board the ship, and US military planes had been launched to support the effort.
However, at the last minute, military officials started looking further into the intelligence report and discovered that it had been created with an AI chatbot, which had erroneously identified the materials on the ship as components of a nuclear weapons program. The resulting report was “entirely false,” one source told CNN.
It is unclear exactly what type of AI model was used in this incident, but CNN reports that a military analyst “queried a chatbot about some intelligence reporting on the ship’s manifest that originated with US Special Operations Command Pacific, based in Hawaii.” The chatbot then reportedly compiled information from open source intelligence (OSINT) and classified signals intelligence (SIGINT) sources, and delivered its erroneous conclusion about what the ship was carrying.
The analyst then reportedly also used AI to package that information in the style of a standard military intelligence report before handing it off for dissemination.
A Heated Debate Over AI Safety
The revelation of the near-catastrophe comes amidst an escalating debate over AI safety, particularly in military contexts.
The Trump administration’s rapid expansion of AI use in the military has generated significant controversy over how quickly the technology is being pushed into systems involving surveillance, targeting, and autonomous weapons before oversight and safety frameworks have caught up — or even been fully developed. There are few clear rules and standards for autonomous AI use, and even when a human is involved in the process, there is “no real guidance” for how having a human in the loop will prevent potentially catastrophic mistakes, a source told CNN.
In January 2026, the Trump administration’s Department of War released a departmental AI strategy that established a mandate for the military to accelerate the use of AI by “aggressively” removing barriers and focusing on speed and scale over safety. Then, in June 2026, Trump released a national security memorandum ordering the adoption of frontier AI models across military and intelligence agencies, and mandating the termination of government contracts with AI developers that attempt to restrict how government agencies use their technology.
The Trump administration claims that if the US doesn’t prioritise the rapid development of military AI capabilities — or if it even slows down to ensure the safety of those systems before deploying them — then China will outpace the US and erode the American military’s technological advantage. But in order to stay ahead in the race against China, the administration has increasingly framed safety as an impediment and has cast those who voice concerns about the speed of AI development as critics trying to stifle American innovation.
Trump Went to War In Order to Break Iran and Left It More Powerful Than Before – US Defence Analyst
Brynn Tannehill explains how Trump’s war has resurrected an Iranian regime that sanctions and drought had been on course to topple
Throwing Caution To The Wind
In a series of social media posts just this past week, the DoW even went as far as portraying the AI safety movement as un-American, and in a bizarre post on Truth Social, Trump said that “the only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ)! PRESIDENT.”
But that’s not what independent experts say.
Human Rights Watch has concluded that military AI adoption is outpacing testing and evaluation, and the Government Accountability Office (GAO) found that the Pentagon’s AI testing policies contain significant gaps and do not require all recommended safeguards, raising serious concerns that problems with these systems may not be identified until they become operational in high-stakes settings — concerns that appear to have already been realised.
There have now been numerous incidents that indicate that the US Government is relying on AI systems that lack proper safeguards and produce information and materials that are unreliable and inaccurate. Some of these incidents – like the State Department’s use of an AI-generated map of Africa that mislabeled numerous countries – have been merely embarrassing, while others – like the bombing of a girls’ school in Iran – have resulted in devastating consequences, including the loss of innocent lives.
We have been told that we must tolerate the extreme risks associated with rapid AI development because if the US ever gets into a war with China, we need to ensure that the American military has more advanced AI weapons than our adversary, but the latest revelations demonstrate that the same technology being marketed as the only way to defeat China militarily may also be the very thing that starts such a war in the first place.