2 hours ago
3
While the artificial intelligence boom is upending sections of the music industry, voice generating bots are also becoming a boon to another unlikely corner of the internet: extremist movements that are using them to recreate the voices and speeches of major figures in their milieu, and experts say it is helping them grow.
“The adoption of AI-enabled translation by terrorists and extremists marks a significant evolution in digital propaganda strategies,” said Lucas Webber, a senior threat intelligence analyst at Tech Against Terrorism and a research fellow at the Soufan Center. Webber specializes in monitoring the online tools of terrorist groups and extremists around the world.
“Earlier methods relied on human translators or rudimentary machine translation, often limited by language fidelity and stylistic nuance,” he said. “Now, with the rise of advanced generative AI tools, these groups are able to produce seamless, contextually accurate translations that preserve tone, emotion, and ideological intensity across multiple languages.”
On the neo-Nazi far-right, adoption of AI-voice cloning software has already become particularly prolific, with several English-language versions of Adolf Hitler’s speeches garnering tens of millions of streams across X, Instagram, TikTok, and other apps.
According to a recent research post by the Global Network on Extremism and Technology (GNet), extremist content creators have turned to voice cloning services, specifically ElevenLabs, and feed them archival speeches from the era of the Third Reich, which are then processed into mimicking Hitler in English.
Neo-Nazi accelerationists, the kinds who plot acts of terrorism against western governments to provoke a societal collapse, have also turned to these tools to spread more updated versions of their hyper-violent messaging. For example, Siege, an insurgency manual written by American neo-Nazi and proscribed terrorist James Mason that became the veritable bible to organizations like the Base and the now-defunct Atomwaffen Division, was transformed into an audiobook in late November.
“For the last several months I have been involved in making an audiobook of Siege by James Mason,” said a prominent neo-Nazi influencer with a heavy presence on X and Telegram, who stitched together the audiobook with the help of AI tools.
“Using a custom voice model of Mason, I re-created every newsletter and most of the attached newspaper clippings as in the original published newsletters.”
The influencer lauded the power of having Mason’s writing from “pre-internet America” and turning it into a modern-day voice.
“But to hear the startling accuracy of predictions made through the early eighties really puts a milestone on the road and it changed my view of our shared cause on a fundamental level,” he said.
At its height in 2020, the Base held a book club on Siege, which was an instrumental influence on several members who discussed its benefits in a hypothetical war against the US government. A nationwide FBI counterterrorism probe eventually swept up over a dozen of its members on various terrorism related charges in the same year.
“The creator of the audiobook has previously released similar AI content; however, Siege has a more notorious history,” said Joshua Fisher-Birch, a terrorism analyst at the Counter Extremism Project, “due to its cultlike status among some in the online extreme right, promotion of lone actor violence, and being required reading by several neo-Nazi groups that openly endorse terrorism and whose members have committed violent criminal acts”.
Webber says pro-Islamic State media outlets on encrypted networks are currently and actively “using AI to create text-to-speech renditions of ideological content from official publications”, to supercharge the spread of their messaging by transforming “text-based propaganda into engaging multimedia narratives”.
Jihadist terrorist groups have found utility in AI for translations of extremist teachings from Arabic into easily digestible, multilingual content. In the past, American-imam turned al-Qaeda operative Anwar al-Awlaki, would personally have to voice English lectures for recruitment propaganda in the anglosphere. The CIA and FBI have repeatedly cited the influence of al-Awlaki’s voice as a key contagion in the spread of al-Qaeda’s message.
On Rocket.Chat – the preferred communications platform of the Islamic State, which it uses to communicate with its followers and recruits – a user posted a video clip in October with slick graphics and Japanese subtitling, remarking on the difficulties of doing that without the advent of AI.
“Japanese would be an extremely hard language to translate from its original state to English while keeping its eloquence,” said the pro-Islamic State user. “It should be known that I do not use artificial intelligence for any related media, with some exceptions regarding audio.”
So far, not just the Islamic State, but groups across the ideological spectrum, have begun using free AI applications, namely OpenAI’s chatbot, ChatGPT, to amplify their overall activities. The Base and adjacent groups have used it for the creation of imagery, while also acknowledging, as far back as 2023, the use of these tools to streamline planning and researching.
Counterterrorism authorities have always viewed the internet and technological advancements as a persistent game of catch-up when it comes to keeping pace with the terror groups who exploit them. Already the Base, the Islamic State and other extremists have leveraged emergent technologies like crypto to anonymously fundraise and share files for 3D printed firearms.
