Artificial intelligence systems are increasingly being exploited by violent extremists for radicalization, weapon planning, and propaganda, according to a report by the non-profit group Tech Against Terrorism. The findings, which categorize the trend as a major national security concern, highlight how generative AI models can function as interactive mentors for individuals seeking to commit acts of violence.
The Shift from Manuals to AI Mentorship
The core danger of modern AI lies in its ability to engage in conversational, iterative dialogue. Adam Hadley, executive director of Tech Against Terrorism, warns that these systems behave less like static manuals and more like personal coaches. Unlike a simple web search, which provides raw data, an AI model can refine dangerous instructions based on a user’s follow-up questions.
In a test of 27 AI models, Tech Against Terrorism found that ChatGPT, the most widely used system, refused only 48 per cent of the more than 2,000 requests entered by testers, such as instructions for 3D-printed firearms or explosive devices. While OpenAI, the developer of ChatGPT, stated that it works “continuously” to improve safety protocols, the report argues that open-source models allow users to remove the safeguards altogether.
Did you know?
The Tech Against Terrorism report documented more than 30 cases where AI systems were used to support terrorist or extremist attacks, resulting in more than 70 deaths.
Escalating National Security Risks
The threat of AI-driven radicalization is no longer theoretical. In Queensland, a 13-year-old boy was recently charged after allegedly using an AI platform to simulate a mass shooting, including requests for a narrative based on real-world violent events. Similarly, authorities have linked a stabbing attack in a Melbourne shopping center to AI-assisted radicalization.
Government officials are taking note of this “step change” in extremist capability. ASIO director-general Mike Burgess warned at the 2024 National Press Club that offshore extremists are already leveraging AI to plan attacks and manufacture weapons. Prime Minister Anthony Albanese recently announced a new national artificial intelligence framework, citing the urgent need to establish a “social licence” for the technology as extremists and state actors turn to AI for recruitment and propaganda.
Regulatory Gaps and Calls for Stricter Enforcement
Current legislative frameworks are struggling to keep pace with the rapid deployment of AI. Under Australia’s Online Safety Act, platforms are required to block pro-terror material, with non-compliance potentially resulting in fines of up to $54 million. However, experts argue these penalties are insufficient for companies with massive global turnovers.
Toby Walsh, chief scientist at the University of New South Wales AI Institute, suggests that regulators should consider more aggressive measures, such as calculating fines based on a percentage of company revenue or imposing geoblocks on platforms that fail to meet safety standards. “I can’t understand why we’re not more outraged,” Dr. Walsh said, noting that if companies cannot follow local laws, their ability to operate within the country should be revoked.
Pro Tip:
Counterterrorism authorities are shifting focus toward the “low barrier to entry” AI creates for younger, impulsive users who may be more susceptible to radicalization through conversational AI interfaces.
Frequently Asked Questions
- How are extremists using AI?
- Extremists use AI to generate propaganda, recruit new members, and receive step-by-step instructions for building weapons or planning violent acts.
- Why is AI more dangerous than a standard internet search?
- According to Tech Against Terrorism, AI models are designed to be persuasive and conversational, acting as a coach that can iterate and refine dangerous information rather than just providing a static link.
- What is the current legal status of AI safety in Australia?
- Australia’s Online Safety Act requires platforms to block pro-terror content. However, there is ongoing debate about whether current fines are large enough to deter major tech companies from distributing unsafe models.
Have you encountered concerns regarding AI safety in your community? Share your thoughts in the comments below or subscribe to our newsletter for the latest updates on technology and national security.
