Jihadist networks are targeting India by shifting from predominantly Kashmir-centric, Urdu-language long-form content toward Hindi and regional-language short-form videos on platforms such as Instagram, YouTube, and WhatsApp. This localisation strategy uses algorithmic amplification and domestic grievances to increase its reach and widen its target audience. The uptick in Hindi content is significant because only around 4–5% of India’s population reports Urdu as a mother tongue or additional language, while Hindi is far more widely understood, allowing the content to reach a much larger mainstream audience.
This Insight draws on India’s National Investigation Agency (NIA) inquiries and open-source data from 2025 to mid-2026. It examines the mechanisms of linguistic content and adaptation, reviews representative cases, and identifies major structural gaps in the current moderation, particularly English-centric and keyword-based approaches, as well as in related counterterrorism systems. It argues that English-dominant monitoring is insufficient for dealing with linguistically localised, short-form ecosystems. It proposes targeted improvements for technology platforms and policymakers. This linguistic shift allows exposure through constant digital consumption rather than deliberate searches, amplifying reach to urban youth and communities outside conventional recruitment pools. Standard monitoring tools, usually calibrated to English or high-profile keywords, struggle to detect these Hindi and regional-language short-form content flows at scale.
In the past, Indian agencies have tracked jihadist propaganda regarding Kashmir, mainly through long-form Urdu materials produced by groups including Lashkar-e-Taiba (LeT), Jaish-e-Mohammed (JeM), Al-Qaeda in the Indian Subcontinent (AQIS), and Islamic State Khorasan Province (ISKP). While short-form jihadist content has circulated in India for over a decade, from 2025 onward, the propaganda has evolved significantly through more systematic translation into Hindi and regional languages and has been optimised for mainstream algorithmic platforms such as Instagram Reels and YouTube Shorts. This Insight examines what factors drive the localisation and mainstreaming of jihadist propaganda in Indian digital spaces and analyses its implications for technology platforms and counterterrorism efforts.
Mechanisms of Localisation and Mainstreaming
Three interconnected factors shape the adaptation of Jihadist content in India. First, linguistic and cultural localisation. Propagandists reframe core jihadist narratives in Hindi and regional languages such as Malayalam, Tamil, Telugu and Kannada, which usually portray Indian Muslims as facing systematic persecution, citing domestic issues including communal tensions, such as the Citizenship Amendment Act, and events in militarised Kashmir to create emotional urgency. Organised media wings such as ISKP’s Al-Azaim Foundation and Al-Isabah, as well as Al-Qaeda’s Islamic Translation Center, have consistently produced or commissioned translations, increasingly assisted by AI tools. Volunteer networks also spread the material with open-source reporting documents Al-Isabah content in Tamil, Malayalam, Hindi and other languages, Al-Azaim material targeting Indian audiences. In some cases, the groups have issued direct calls for content to be spread in Indian languages to reach wider audiences. This allows propagandists to combine global jihadist ideology with everyday Indian realities, making the content feel less foreign and more personally relevant to users who may have no prior exposure to traditional extremist material.
The next stage is format optimisation. Extended lectures that are traditionally long-form Urdu ideological talks and official media releases produced by groups such as Lashkar-e-Taiba, Jaish-e-Mohammed, Al-Qaeda in the Indian Subcontinent (AQIS), and earlier Islamic State media arms have now largely been replaced by concise, emotionally driven short videos. These clips typically feature dramatic visuals, text overlays in Hindi or regional languages, fast-paced editing, and nasheeds.
These formats align closely with platform recommendation systems that prioritise high engagement metrics such as watch time, shares, and comments. While general research has examined how recommendation algorithms can amplify extremist content, there is currently no publicly available granular data that measures the algorithmic performance of jihadi short-form content specifically in the Indian context. The clips themselves are typically short and designed for rapid consumption on platforms. Open-source monitoring and Indian intelligence assessments describe them as visually intense, usually featuring fast-paced editing, dramatic imagery, text overlays in Hindi or regional languages, and emotionally charged soundtracks. Because precise comparative statistics on organic reach are not publicly available, intelligence assessments and open-source reporting suggest that these localised short-form videos circulate faster and wider than traditional long-form Urdu lectures, which require sustained attention.
Finally, once the content is created, it is spread across a layered digital ecosystem. Initial discovery often occurs on open, recommendation-driven platforms such as Instagram and YouTube. Users who engage with the material are then frequently moved into closed messaging environments, particularly Telegram groups and, in some cases, WhatsApp channels or groups, for further interaction and ideological reinforcement. NIA investigations into recent online radicalisation networks such as the 2026 multi-state case originating in Vijayawada described a recurring pattern in which contacts begin on open platforms and are subsequently shifted to encrypted apps for deeper engagement with ISIS- or AQIS-linked material. In these closed spaces, individuals are exposed to specific ideological content and are sometimes connected with handlers. JeM has also made use of publicly accessible WhatsApp Channels to distribute ideological material in Hindi and English to Indian audiences. AI-assisted tools are also being employed to speed up translation into Indian languages and generate or adapt content. Indian intelligence assessments have identified AI-generated videos spread in multiple Indian languages for radicalisation purposes. Decentralised supporter networks further regenerate and redistribute material after platforms remove it. Indian intelligence assessments have noticed an increase in volunteer-run channels that amplify ISKP propaganda. Unofficial pro-IS media initiatives with an India focus, such as Al-Jauhar Media, have also created and spread material in Hindi, Malayalam and other languages, aligning with broader Islamic State messaging. This mixture of open-platform discovery, closed-environment radicalisation, technological assistance, and supporter networks forms a very flexible content dissemination system which is resilient and difficult to permanently disrupt.
Jihadist Online Radicalisation in India
NIA operations and open-source reporting from 2025 to mid-2026 illustrate how this model operates in practice and demonstrate the impact of the localisation strategy. In July 2026, the National Investigation Agency conducted coordinated searches across 20 locations in 10 states and Delhi, targeting an ISIS/AQIS-linked online radicalisation network which originated in Vijayawada, Andhra Pradesh. The investigation showcased the central role of social media. One documented case involved Sayeeda Begum, a Hyderabad single mother and former domestic help who reportedly built an Instagram following of over 38,000 users. She allegedly transitioned from lifestyle content (such as cooking and daily life reels) to sharing extremist materials and contact information for jihadi handlers. This case illustrates the mainstreaming pipeline, where everyday lifestyle reels can be used as an entry point to gradually expose users to more radical content. In southern India, intelligence assessments recorded the expanded ISKP volunteer channels targeting the southern Indian States of Kerala, Karnataka, and Andhra Pradesh. These channels spread AI-translated short videos in Hindi and Malayalam that combined local issues, such as economic marginalisation and communal incidents, with broader jihadist framing. The use of regional languages helped the content reach audiences in southern India more effectively than traditional Urdu material.
A March 2026 case in Lucknow involved the arrest of a 19-year-old dentistry student who allegedly used Instagram to identify vulnerable youth and directed them to encrypted platforms (Discord and Session) for progressive exposure to ISIS content. According to the Uttar Pradesh Anti-Terrorism Squad (ATS), the pattern started with religious posts before moving to private group invitations and ideological intensification. Jaish-e-Mohammed has also adapted by using publicly accessible WhatsApp Channels. Channels such as Markaz Sayyedna Tamim Dari (MSTD), which amassed over 13,000 followers, regularly distributed Hindi and English ideological material, including religious lectures mixed with commentary on current Indian events. This approach illustrates a parallel form of platform adaptation that increased the group’s reach to professional and urban audiences through an easily accessible channel.
Taken together, these examples indicate a radicalisation progression that starts with initial exposure through algorithmically recommended short-form content on open platforms, followed by deeper engagement in closed messaging groups and, in some instances, direct connection with handlers. This structured pipeline demonstrates how the localisation strategy has made jihadist propaganda much more effective at reaching diverse demographic groups across India.
Implications and Detection Gaps
This localisation presents many structural challenges for platforms and authorities. Algorithmic systems favour high-engagement short-form content, enabling rapid scaling before human or automated moderation can intervene or slow it down. Multilingual detection capabilities are limited in this case, particularly for Hindi and regional languages, where different contexts, sarcastic wording, and coded language are difficult for current AI systems to interpret accurately, as they are predominantly trained on English and high-resource data. Research on content moderation in Indic languages has consistently shown weaker detection performance and under-resourcing for non-English content. Closed messaging layers such as WhatsApp and Telegram also obscure the later stages of radicalisation, making it difficult for external observers to track progression from exposure to action. Collectively, these factors create a resilient, scalable, and fast pipeline that is poorly addressed by traditional English-centric or keyword-based monitoring approaches.
Recommendations
Technology platforms should prioritise developing and using multilingual, trajectory-based detection systems that can recognise suspicious shifts in user content patterns across sessions and platforms, with specific attention to Hindi and regional languages rather than relying solely on isolated keywords. Such systems should focus on behavioural indicators such as rapid transition from mainstream religious content to extremist framing, increased engagement with grievance-based videos, or movement to private groups. Although fully automated detection of radicalisation trajectories (shifts in a user’s content and platform behaviour over time) remains technically and ethically difficult, especially around definitions of extremism and privacy in closed apps, the existing behavioural-analysis approaches provide useful starting points. Research on behavioural indicators of radicalisation and tools that analyse posting patterns over time provide models that could be adapted for Hindi and regional-language contexts, if they are developed with human oversight and clear proportionality guardrails. Platforms should also improve cross-platform collaboration, based on mechanisms such as the Global Internet Forum to Counter Terrorism (GIFCT) hash-sharing database, which allows members to share digital fingerprints of known terrorist and violent extremist content. This would also help identify signal-sharing on emerging linguistic trends in Indian languages and improve collective response capacity, without curbing any platform’s independent enforcement decisions.
Platforms should also establish clearer, jurisdiction-sensitive referral protocols with Indian agencies to improve timely response while balancing privacy safeguards and proportionality. These protocols must pay explicit attention to privacy safeguards, necessity, and proportionality. This is important as the localisation of jihadist propaganda into regional languages, along with the quick movement of users from open platforms into closed environments, creates detection and response gaps that purely automated or English-centric systems are poorly equipped to address. There is a need for structured referral mechanisms calibrated to Indian legal frameworks. Without them, platforms risk either under-responding to credible threats or sharing information in a manner that lacks necessity, transparency, and proper privacy protection.
They must recognise the legal and practical difficulties of accessing content on encrypted platforms and should be designed around clear margins for referral, independent governance and respect for both user privacy and national security requirements. Using existing models of cross-platform cooperation, such as GIFCT, and adapting them to the Indian jurisdictional context would help make consistent referrals that are rights-respecting and operationally useful. Greater investment in regionally trained content moderators and AI models customised for Indian linguistic and cultural contexts would improve detection efforts. For policymakers and PCVE practitioners, priorities include building language-capable open-source intelligence for threat assessment and raising awareness of short-form propaganda dynamics in community-level digital literacy initiatives. Enhanced, transparent cross-platform information sharing on emerging linguistic trends would also help in mitigating this threat. These recommendations are necessarily incremental. Definitions of extremism remain contested, access to content on encrypted platforms raises major privacy and legal hindrances, and current detection systems lack linguistic nuance and behavioural context. Any impactful response will need to proceed carefully, building on existing tools and cooperation mechanisms while remaining realistic about their limits.
–
Ashreet Acharya is a Lecturer in Law at Liverpool Hope University. His research focuses on conspiracies and political extremism. He holds master’s degrees in Intellectual Property Rights and Data Protection and Corporate Laws and is an Indian bar-qualified Advocate. Ashreet has diverse experience in academia and the ed-tech sector, with several publications on contemporary legal and social challenges.
–
Are you a tech company interested in strengthening your capacity to counter terrorist and violent extremist activity online? Apply for GIFCT membership to join over 30 other tech platforms working together to prevent terrorists and violent extremists from exploiting online platforms by leveraging technology, expertise, and cross-sector partnerships.
The views and opinions expressed in this Insight are the authors’ own and do not necessarily reflect those of GNET or any of its partners.