IS turns to artificial intelligence for advanced propaganda amid territorial defeats
A new form of propaganda developed by IS supporters is broadcasting news bulletins with AI-generated anchors in multiple languages.
The Islamic State Khorasan (ISKP) group, an IS affiliate active in Afghanistan and Pakistan, produced in a video an AI-generated anchorman to appear reading news following an IS-claimed attack in Bamiyan province in Afghanistan on May 17 that killed four people, including three Spanish tourists.
The digital image posing as an anchor spoke the Pashto language and had features resembling local residents in Bamiyan, according to The Khorasan Diary, a website dedicated to news and analysis on the region.
Another AI-generated propaganda video by Islamic State appeared on Tuesday with a different digital male news anchor announcing IS’s responsibility for a car bombing in Kandahar, Afghanistan.
“These extremists are very effective in spreading deepfake propaganda,” said Roland Abi Najem, a cybersecurity expert based in Kuwait.
He told VOA that a group like IS was already effective in producing videos with Hollywood-level quality, and the use of AI has made such production more accessible for them.
“AI now has easy tools to use to create fake content whether it’s text, photo, audio or video,” Abi Najem said, adding that with AI, “you only need data, algorithms and computing power, so anyone can create AI-generated content from their houses or garages.”
IS formally began using the practice of AI-generated news bulletins four days after an attack at a Moscow music hall on March 22 killed some 145 people. The attack was claimed by IS.
In that video, IS used a “fake” AI-generated news anchor talking about the Moscow attack, experts told The Washington Post last week.
Mona Thakkar, a research fellow at the International Center for the Study of Violent Extremism, said pro-IS supporters have been using character-generation techniques and speech-to-text AI tools to produce translated news bulletins of IS's Amaq news agency.
“These efforts have garnered positive responses from other users, reflecting that, through future collaborative efforts, many supporters could produce high quality and sophisticated AI-powered propaganda videos for IS of longer durations with better graphics and more innovation techniques,” she told VOA.
Thakkar said she recently came across some pro-IS Arabic-speaking supporters on Telegram who were recommending to other supporters “that beginners use AI image generator bots on Telegram to maintain the high quality of images as the bots are very easy and quick to produce such images.”
AI-generated content for recruitment
While IS’s ability to project power largely decreased due to its territorial defeat in Syria and Iraq, experts say supporters of the terror group believe artificial intelligence offers an alternative to promote their extremist ideology.
“Their content has mainly focused on showing that they’re still powerful,” said Abi Najem. “With AI-generated content now, they can choose certain celebrities that have influence, especially on teenagers, by creating deepfake videos.”
“So first they manipulate these people by creating believable content, then they begin recruiting them,” he said.
In a recent article published on the Global Network on Extremism and Technology, researcher Daniel Siegel said generative AI technology has had a profound impact on how extremist organizations engage in influence operations online, including the use of AI-generated Muslim religious songs, known as nasheeds, for recruitment purposes.
“The strategic deployment of extremist audio deepfake nasheeds, featuring animated characters and internet personalities, marks a sophisticated evolution in the tactics used by extremists to broaden the reach of their content,” he wrote.
Siegel said that other radical groups like al-Qaida and Hamas have also begun using AI to generate content for their supporters.
Cybersecurity expert Abi Najem said he believes the cheap technology will increase the availability of AI-generated content by extremist groups on the internet.
“While currently there are no stringent regulations on the use of AI, it will be very challenging for governments to stop extremist groups from exploiting these platforms for their own gain,” he said.
This story originated in VOA’s Kurdish Service.