UK deploys AI to prevent military secrets leaks

The British military is reportedly turning to artificial intelligence to reduce the risk of data leaks due to human error.

The UK Ministry of Defence has selected technology from Australian company Castlepoint Systems to help reduce accidental disclosures and improve the handling of sensitive documents.

AI tool will help the British army control military secrets. Photo: Explainable

Under the initiative, the company's Explainable AI tool will be deployed to scan documents and emails on military networks in real time, flag sensitive content, and apply automated safeguards to minimize errors when managing large data sets.

Unlike traditional tools that only detect simple data points like names or credit card numbers, the AI engine will use rules-based analysis to classify complex content, including references to explosives or military locations.

The tool can also review old records with missing or outdated labels, adjust classifications as needed, and update markers to ensure sensitive data is still properly protected.

A list of thousands of Afghans who cooperated with the British government was once leaked due to a mistake by a soldier. Photo: The Guardian

“Recommendations are fully traceable to specific matches, tailored to an organization’s unique risk profile. This makes the classification process transparent, explainable, and debateable, meeting the highest standards of ethical AI,” the company claims.

The move follows an incident in 2022 when a British soldier accidentally sent a spreadsheet containing personal information about members of the Special Air Service and the Secret Intelligence Service to contacts in Afghanistan involved in resettlement programmes.

According to The Guardian , the document includes more data than expected, including the names of British officials and thousands of Afghans seeking resettlement after the Taliban takeover.

British news outlets reported that the breach was one of 49 security incidents recorded by London in the past four years, although only four were made public.

While Explainable AI aims to minimize the risk of leaks due to human error, cybersecurity officials have warned that the expanded use of AI could create new vulnerabilities and advise against rapid adoption without safeguards.