As the rest of the world rushes to harness the power of artificial intelligence (AI), extremist groups are also experimenting with the technology, even though they don't know exactly what to do with it.
AI could become a powerful tool for extremist groups to recruit new members, create large numbers of realistic deepfake images and refine cyberattacks, national security experts and spy agencies have warned.
Last month, someone posting on a pro-Islamic State group's website urged other Islamic State supporters to make AI part of their efforts. “One of the best things about AI is its ease of use,” the user wrote in English.
“Some intelligence agencies are concerned about the contribution AI will make.” [to] “So make their nightmares a reality,” the user continued.
The Islamic State, which captured territory in Iraq and Syria years ago and is now a decentralized alliance of armed groups that share a violent ideology, realized years ago that social media could be a powerful tool for recruitment and disinformation, so it's no surprise that the group is experimenting with AI, national security experts say.
Whether it's a tightly knit, poorly resourced extremist group, or an individual villain with a web connection, AI can be used to push out propaganda and deepfakes at scale, expanding their reach and increasing their influence.
“AI makes things a lot easier for any adversary,” said John Laliberte, a former National Security Agency vulnerability researcher and current CEO of cybersecurity company ClearVector.
“With AI, even small groups that don’t have a lot of money can have an impact.”
How extremist groups conduct experiments
As soon as programs like ChatGPT became widely accessible, extremist groups began using AI. Since then, they have increasingly used generative AI programs to create lifelike photos and videos.
When manipulated by social media algorithms, this fake content can help gather new followers, confuse and scare opponents, and spread propaganda on a scale unimaginable just a few years ago.
Two years ago, these groups spread fake images from the Israeli-Hamas war depicting a bloody baby abandoned inside a bombed building. These images fostered anger and polarization while obscuring the real horrors of the war. Violent groups in the Middle East used the photos to recruit new members, as did anti-Semitic hate groups in the United States and elsewhere.
A similar incident occurred last year after an attack by Islamic State affiliates claimed to have killed nearly 140 people at a concert venue in Russia. In the days after the shooting, an AI-generated propaganda video was widely circulated on bulletin boards and social media to recruit new employees.
The Islamic State also created deepfake audio recordings of its leaders reciting scripture and used AI to quickly translate messages into multiple languages, according to researchers at Site Intelligence Group, which tracks militant activity and examines IS's evolving use of AI.
“Ambitious” — for now
Marcus Fowler, a former CIA official and now CEO of Darktrace Federal, a cybersecurity firm that works with the federal government, said these groups lag behind China, Russia and Iran, and still see more advanced uses of AI as “ambitious.”
But the risks are too high to ignore and are likely to increase as the use of cheap and powerful AI expands, he said.
Hackers are already using synthetic voice and video in phishing campaigns, impersonating senior corporate and government leaders to gain access to sensitive networks. AI can also be used to write malicious code and automate some aspects of cyberattacks.
Of further concern is the possibility that armed groups may seek to use AI to assist in the production of biological or chemical weapons to compensate for a lack of technical expertise. This risk was included in the Department of Homeland Security's latest Homeland Threat Assessment released earlier this year.
“ISIS got into Twitter early and found a way to use social media to their advantage,” Fowler said. “They're always looking for the next thing to add to their arsenal.”
Combat growing threats
Lawmakers have proposed several proposals that they say require urgent action.
For example, Sen. Mark Warner of Virginia, the top Democrat on the Senate Intelligence Committee, said the U.S. needs to make it easier for AI developers to share information about how their products are being used by bad actors, including extremists, criminal hackers and foreign spies.
“Since late 2022, it has become clear that the same enthusiasm and experimentation with generative AI that the public has had since ChatGPT became publicly available applies to a wide variety of malicious actors,” Warner said.
During a recent hearing on extremist threats, House members learned that the Islamic State and al-Qaeda are holding training workshops to help their supporters learn how to use AI.
A bill passed by the U.S. House of Representatives last month would require Homeland Security officials to annually assess the AI risks posed by these groups.
Preventing malicious use of AI is no different than preparing for more traditional attacks, said August Pflueger, a Texas congressman and sponsor of the bill.
“Our policies and capabilities must respond to tomorrow's threats,” he said.
