TLDR: Pro-ISIS outlets are actively recommending and leveraging encrypted generative AI services to produce and disseminate propaganda, circumvent content moderation, and bolster recruitment. While a specific new service named ‘Lumo’ has reportedly been endorsed, detailed information regarding this particular platform was not found in recent counter-terrorism analyses. The broader trend indicates a significant shift in extremist groups’ digital tactics, utilizing AI for enhanced visual content, personalized messaging, and operational planning.
Recent reports from leading counter-terrorism organizations highlight a growing trend among pro-ISIS channels: the endorsement and active use of encrypted generative AI services for various illicit activities. While a specific new encrypted generative AI service named ‘Lumo’ has reportedly been recommended by a pro-ISIS outlet, detailed information on this particular platform was not available in the comprehensive analyses reviewed. However, the broader landscape of AI exploitation by violent non-state actors (VNSAs), including the Islamic State (ISIS), is well-documented.
According to a February 2024 report titled ‘AI Caliphate: The Creation of Pro-Islamic State Propaganda Using Generative AI’ by Meili Criezis for GNET (Global Network on Extremism & Technology), terrorist organizations are increasingly exploiting artificial intelligence for a range of purposes, including recruitment, propagandizing, online account preservation, communication, and even attack planning . A 2023 Tech Against Terrorism report, cited in the GNET study, revealed that a ‘pro-IS tech support group’ had already shared a guide on utilizing AI content generators, and pro-IS supporters were employing AI for transcribing leadership messages and creating or enhancing visual propaganda . The study analyzed 286 pieces of AI-generated or enhanced propaganda from pro-IS accounts across platforms like Instagram, Meta, Pinterest, and Pixiv.
A prominent figure in this digital landscape, referred to pseudonymously as ‘IslamicState_Supporter,’ was identified as an original creator actively integrating AI into their propaganda production. This individual’s content, often combining AI capabilities with hand-drawn art and photo editing software, showcased themes such as ISIS flags, weapons, solitary mujahids, and explicit depictions of violence, including executions and terrorist attacks . The prevalence of ISIS flags in 45% of posts, despite the risk of account bans, underscores the importance placed on promoting the ‘IS brand’ .
These propagandists are also employing sophisticated content moderation evasion strategies. For instance, they blur ISIS flags or overlay stickers on content shared on platforms like Instagram and Meta, while simultaneously distributing unaltered versions on less-moderated sites such as Pinterest and Pixiv . Interestingly, some AI-generated content faced criticism from fellow pro-IS supporters who deemed it ‘un-Islamic’ or ‘haram’ if it featured faces, leading some creators to adapt by focusing solely on flags .
Further insights from an April 2025 policy brief by the Global Internet Forum to Counter Terrorism (GIFCT) and the Konrad-Adenauer-Stiftung (KAS), titled ‘Artificial Intelligence: Threats, Opportunities, and Policy Frameworks for Countering VNSAs,’ corroborate these trends . The report emphasizes that VNSAs are now able to ‘more easily scale the quality and quantity of propaganda and TVEC (Terrorist and Violent Extremist Content) through GenAI products’ .
Key concerns highlighted include:
Content Moderation Evasion: AI tools are used to manipulate TVEC to bypass platform detection, such as overlaying cartoon characters on livestreamed attack footage .
Propaganda and Disinformation: GenAI facilitates the creation of contextually tailored and personalized propaganda, including ‘deepfakes’ of individuals and audio re-creations. ‘More refined GenAI images are now seen across official ISIS-affiliated publications’ .
Recruitment and Radicalization: AI strengthens personalized recruitment efforts, with the potential for ‘recruitment chatbots’ in social media and gaming environments, and ‘deepfakes’ of attackers inciting violence .
Attack Planning and Operationalization: AI tools can provide critical information for planning attacks (e.g., bomb-making, 3D printed firearms, chemical/biological weapon design) and there are fears of VNSAs utilizing AI-enhanced drones or autonomous vehicles for attacks .
Also Read:
- AI’s Rapid Evolution: Misinformation, Cyber Threats, and Market Shifts Dominate Recent Headlines
- Generative AI: A Crucial Tool in the Evolving Battle Against Financial Crime
In response, the tech sector is implementing mitigation efforts, including prompt filtering to evaluate user inputs and contextual guardrails to detect patterns indicating attempts to ‘jailbreak’ AI models . Both reports underscore the critical need for tech companies to identify specific keyword combinations used by extremists to generate harmful content and to address existing gaps in content moderation strategies. The evolving nature of this threat necessitates robust multi-stakeholder collaboration and continuous adaptation of counter-terrorism measures in the digital realm.
