Media: UK security experts alarmed as ISIS-K trains recruits in AI use

British experts have warned that the Islamist group Islamic State Khorasan (IS-K) is actively promoting the use of artificial intelligence (AI) among its recruits, a development that could alarm lawmakers and security agencies, POLITICO reports.

The group’s English-language magazine, Voice of Khorasan, has devoted pages in its two most recent editions to instructing supporters on how to use AI to be a “responsible mujahid.” IS-K, the Afghanistan branch of ISIS, has been linked to attacks in Russia, Iran, Afghanistan, Pakistan, and the foiled Taylor Swift concert attack in Vienna in 2024. Last year, the European Union added the magazine’s publisher, the Al Azaim Media Foundation, to its sanctions list.

The latest edition of the magazine features a section headed with the warning: “AI is like fire. You can use it to light a home, or to burn it down.” The magazine offers advice on “responsible ways to use AI chatbots,” including assisting in religious campaigns and preaching. It also advises supporters to use AI for anonymous private research that avoids “sharing sensitive information” and “exposing ourselves to unnecessary risks.”

The magazine cautions that chatbots should not be used for sharing personal information, uploading confidential files, submitting political or security-sensitive queries, or issuing religious rulings. A previous edition compares AI models from OpenAI, Anthropic, Meta, Microsoft, and DeepSeek, warning against tools that might involve “Israeli Defense Forces AI infrastructure” and recommending the privacy-focused browser Brave Leo for “sensitive queries.”

The publication also frames AI literacy as a religious obligation, describing it as a “fard al-ayn” — an individual duty akin to prayer, fasting, or pilgrimage. One issue states: “AI is no longer optional, it's your shield and compass in a digital world wired with hidden threats.”

Jonathan Hall, the UK government’s independent reviewer of terrorism legislation, told POLITICO that these developments highlight the accelerating risk of AI being exploited by terrorist groups. “The developments in AI and agentic AI are accelerating fast, and more and more of these capabilities are available off the shelf to terrorist groups,” he said.

“LLMs are lowering the barriers to creating sophisticated and tailored propaganda, which is a problem as terrorist groups increasingly seek to exploit local grievances in local languages. I would not be surprised if chatbot radicalisation starts to take off: If you can create a terrorist website, why would you refrain from creating a terrorist chatbot? If government officials are not watching a ticker-tape of AI developments, they really should be."

Avi Jager, senior director of the Severe Harms Intelligence Department at AI security company Alice, noted that the instructions reflect not only “tactical guidance” but an “ideological shift with significant long-term implications.” He explained that in the past, jihadist movements were often wary of emerging technologies like AI, with some arguing that using such tools might be un-Islamic or contradictory to divine guidance. “That’s what makes this moment so noteworthy: ISIS-K is no longer just tolerating artificial intelligence, it is explicitly endorsing its use under certain conditions,” Jager said.

By Vafa Guliyeva