Propaganda and Terror: UN Reveals AI’s Role in Terrorist Strategies
Terrorist propaganda is no longer limited to empty rhetoric. It has become a direct instrument for recruitment and radicalisation, powered by modern technology and designed to expand the reach of fear.
This is the conclusion of a report prepared by a United Nations committee pursuant to UN Security Council resolutions 2617 (2021) and 2810 (2025). The document was published in September alongside the meetings of the UN General Assembly.
According to the report, terrorist groups use propaganda as a core operational tool for recruitment, incitement to violence, fundraising and the provision of guidance to their followers.
Although terrorists have long made propaganda an integral part of their strategies, both its methods and scale have evolved considerably. Its function is now more central, integrated and adaptable than in earlier periods.
Contemporary terrorist propaganda therefore possesses distinctive characteristics that shape not only how it is produced but also how it operates in practice in relation to recruitment, influence and strategic communication.
These features determine how propaganda attracts individuals by embedding ideological narratives within their everyday experiences and perceptions of injustice or grievance.
Artificial Intelligence
Successive UN reports have documented a growing pattern in the use of artificial intelligence by terrorist organisations.
These groups are using AI, particularly generative systems, to strengthen their propaganda activities, primarily for recruitment and radicalisation. They have also issued guidance to supporters on how to avoid detection while using such tools.
According to the report, artificial intelligence has the potential to narrow the capability gap between terrorist organisations and state-led security systems.
Terrorist groups are increasingly turning to AI to enhance the precision and intensity of attacks and to incite violence.
The technology can be used to produce and distribute propaganda more quickly and efficiently than ever before. Its applications range from interactive recruitment and behavioural influence through social media channels to the dissemination of disinformation and ideologies that encourage terrorism.
AI-powered bots can amplify violent extremist material on a large scale, making it more difficult to detect and counter.
At the same time, they allow potential recruits to be targeted with personalised messages tailored to their interests, beliefs and grievances.
AI applications may also unintentionally accelerate the radicalisation of vulnerable individuals, the report warns, as users increasingly rely on such systems for companionship, guidance and advice.
Recent research suggests that artificial intelligence may function not merely as a tool but as an environment for radicalisation.
Continuous, personalised interactions could reinforce existing grievances, identities and beliefs while remaining largely invisible to traditional counterterrorism methods.
Artificial intelligence may also strengthen the gender dimension of terrorist propaganda.
Experts have observed that generative AI enables the rapid production and adaptation of gender-targeted propaganda for men, women, girls and boys.
This allows terrorist and violent extremist actors to tailor their narratives to specific audiences on a wider scale and distribute them at greater speed.
A Troubling Development
Specific operational uses have been documented across different groups and regions.
Supporters of the Islamic State group launched an AI-generated programme called “News Harvest” to distribute propaganda videos.
More broadly, the report says terrorist actors are demonstrating increasing sophistication in integrating artificial intelligence tools and visual effects into propaganda content.
They are also using AI-enhanced translation to enable rapid, simultaneous distribution across multiple languages.
The report further notes that the Islamic State sought to recruit cyber specialists to strengthen its capabilities and adapt its propaganda to local contexts.
More troublingly, recent research has highlighted shortcomings in the safeguards used by current artificial intelligence systems.
In tests involving 27 leading AI models, approximately one-third reportedly provided information that went beyond what was readily available through a conventional internet search, including assistance related to attack planning or the manufacture of explosive devices.
The research also found that relatively simple changes to prompts could sometimes circumvent safety mechanisms.
These findings indicate that existing safeguards remain inconsistent and vulnerable to misuse.
