Jihadists and AI

0:00This BBC podcast is supported by ads outside the UK. America is changing. And so is the world. But what's happening in America isn't just a cause of global upheaval. It's also a symptom of disruption that's happening everywhere. I'm Asma Khalid in Washington, D.C. I'm Tristan Redman in London. And this is The Global Story. Every weekday, we'll bring you a story from this intersection, where the world and America meet. Listen on BBC.com or wherever you get your podcasts.

0:45I'm very concerned that we are accelerating on a road that we've never been. We could lose control very easily. My worst fear is that we, the industry, cause significant harm to the world. I think if this technology goes wrong, it can go quite wrong. AI could give weapons to terrorists and create new pandemics. It could upend geopolitical stability. We are not equipped to deal with this. Two men who have something to say about the dangers of artificial intelligence. Sam Altman is the creator of ChatGPT, and Yoshua Bengio is described as one of the godfathers of AI. The potential of this tech in the hands of bad actors has also been on the radar of BBC Monitoring's jihadist media team.

1:36IS supporters are very early adopters of tech generally in the past, and a lot of them, particularly the younger ones, are very tech savvy. Sympathisers of the Islamic State group and al-Qaeda have been eyeing the tech for a while. They've had lots of publications encouraging their followers and saying that this is important, you have to learn about AI. There were IS fans saying, why don't we use AI? This is a real boon to us. This is a way we can spread the messaging. We can reach a much younger audience, a much wider audience. So what machine-generated jihadist material is out there? Posters.

2:12The videos. AI for translations. An audio reader. An animated representation of an IS militants. Could AI-powered jihadist propaganda be lurking on your social media feeds? or worse, on those of your children. It was a cartoon and they said, this is also to reach children. We want to inspire young people and kids to learn about the Aya story. In this episode, we get inside the mindset of hardcore supporters of jihadist ideology, weighing up their enthusiasm for this powerful tech against their fears about the damage it could inflict on their image and reputation.

2:57Welcome to the documentary from the BBC World Service. I'm Krasi Ivanova-Tweak, and this is the global jigsaw from BBC Monitoring, where we look at the world through the lens of its media, exploring the narratives, the propaganda and the disinformation. Radio Tehran. We didn't get on the territory of Ukraine. Western countries are not the international community. That era has finished. All right, Steve, was there anything you wanted to say something like? No, I think generally it's still a young technology. In this episode, you'll hear from two members of BBC Monitoring's jihadist media team, Mina Alami and Stephen Humphreys, who'll unpack what responsible AI looks like in the eyes of jihadist supporters.

3:50So far, jihadist use of AI has been inconsistent, somewhat patchy. It seems that currently they are experimenting with the various AI tools, watching, learning and trying to master the use of AI. Some IS and Al Qaeda supporters, as well as specialist media groups have been using AI, mostly for translations. You might have come across some alarming takes in Western media on the potential dangers of an AI-armed jihad. Here's one from the BBC. Here in the UK, a counter-extremist think tank says we need new laws that reflect the danger of AI being used to recruit terrorists, says the BBC. Several bots were found to mimic extremist groups.

4:39The debate made it to the UN as well. Member states reports that Daesh may be seeking to recruit experts in cyber security. Daesh's use of artificial intelligence and social media is being harnessed to amplify the group's reach. It also holds potential for states to enhance the detection of terrorist activities. The threat posed by lone individuals radicalized by ISIL ideology, and these are often young in age, is rising. Assistance like ChatGPT could translate an editorial in seconds, repackage it and tailor it for different audiences around the world. Yet our team has picked up on considerable caution among jihadists when it comes to using the tech in cognitive warfare.

5:28IS's weekly newspaper An Nebbe, which comes out in Arabic every Thursday. The centerpiece of that publication is the editorial, and IS uses it to communicate its key messages to followers, to operatives around the world, to comment on big developments, events in Gaza, incitement. But it's not accessible. It's very complicated. It's full of flowery language. The message is usually hidden in between, but also it's classical Arabic. So it's even for Arabic speakers, not everyone will understand it. But because we were fluent in both languages, we can say for sure that AI was used in recent months to translate issues of a nibet to English specifically, and possibly to other languages, in order to reach a global, mainly English-speaking audience.

6:18So they are courting a wider audience by making the message accessible in more languages. What other AI tricks have they used? It's also been used to enhance dull official texts from IS or Al-Qaeda. They're paired with images, visuals, audio, just to make it more appealing also to a younger audience. Posters. We do see some posters, don't we? And the videos that they've used to repackaging again. But that's not just text or an audio reader. We've actually seen they use archive IS footage and overlay the commentary. Usually in Arabic to Arabic commentary, though, it's not into English. And transcription of some products.

6:58So there would be a video and I can't say for sure that it's AI, but it quickly gets transcribed into kind of a full version of all the images, blow by blow with the narration and all of that. But we know also for sure that they are using AI on social media. So Meta, for example, it does flag up if any AI was used in a post, whether to create content or in translation, it would have an AI tick. and I see multiple posts with visuals on both Instagram and Facebook with the AI tick. So we know that they're using it to enhance and spread IS messaging on these big social media platforms. Have you seen examples of anything else like voice cloning and something far more advanced than photoshopping something on a poster?

7:45A pro-IS media outlet that was pretty unknown to us before started issuing animated news bulletins with a newsreader in Arabic garb. And he'd read out recent official IS attack claims, just voice-to-text transliteration, obviously machines generated. And then they started sharing reports and narratives from the IS's official Naba newspaper. Hesad was the name of this outlet. Before it started publishing its AI news bulletins, There was a debate on a forum that was used by AI supporters, many of whom were subsequently sharing these bulletins. The newsreader Stephen speaks of is not a real person.

8:27It's AI generated. The debate it provoked might explain why it remained an isolated example. We had a small group of users not issuing any ideological objection or technical objection, almost a sort of objection of taste, saying that this is puerile. This is like doing YouTube shorts. Our message is very grave, it's very serious. We are spreading the word of Islam. We do not resort to cheap cartoons and gimmicks. The list of objections is long, not just about style and aesthetics, but rooted in fundamental ideological concepts. This newsreader has his face uncovered. Now this is haram in strict Islam.

9:08You cannot show craven images of animate beings. Isabel's blanked out the face, changed the clothing, because there was also criticism that the garb was not Islamic clothing. So he changed it eventually to a sort of ninja-style, very IS militant-looking character. But even that was not enough to convince the detractors. We have people saying, yeah, but you're using a machine-generated voice that is likely based on the voice of a non-believer. Because it's reading official IS material, which contains a lot of wording from scripture, you cannot have a non-Muslim reading the Holy Scripture. The outlet did seem to shut itself down, but that was not because of criticism.

9:43and that came after his bulletins actually made it into mainstream news reporting. And the terror group has even been experimenting with videos generated by artificial intelligence to spread its hate message... And I think it spooked them to some extent and we haven't seen any similar attempt news bulletins that Hissad did, which is unusual because they did prove quite popular, didn't they? They sort of captured the imagination of a lot of my supporters. There were IS fans saying, why don't we use AI? This is a real boon to us, this is a way we can spread the messaging, we can reach a much younger audience, we can reach a much wider audience.

10:16Once again, we are reminded just how divided jihadists are, not only on the use of AI, but on other issues we've explored in previous episodes. There are several reservations that could be a generational thing. It could be that mature supporters of IS feel that by using these gimmicky videos, we might be trivialising the message, the cause. There was a similar backlash, not necessarily done through AI a couple of years ago. It was telling the story of IS's late spokesman, Abu Muhammad al-Adnani, and it was as a cartoon. And they said, this is also to reach children. We want to inspire young people and kids to learn about the IS story.

10:56And the criticism was that it was trivializing the IS story or the narrative, and it depicted the spokesman as a cartoon figure. And again, they had to kind of blur the face of the spokesman because initially he did actually have facial features, which was forbidden. This has been going on for many years, not even just in relation to AI, in relation to social media platforms. So from what you're telling me, I understand IS are much more keen on this technology than Al-Qaeda. We've seen more of it being used by IS supporters than Al-Qaeda. Al-Qaeda considered the more academic, more mature group and they put out these long essays and lectures and sermons, they're not going to run after the next shiny thing.

11:40So I think there's reservations about that, but they're not against it. I think they're just kind of still trying to find out what is the best way to use AI to present a certain compelling message. What's holding back jihadists isn't just what AI can do for them, It's also what it could do against them. Several pro-Al Qaeda media groups have put out really kind of big documents about the use of AI, mostly taken from mainstream publications in the West. So they would provide translations of what AI is, how it's being used. They want to know, first and foremost, how it's being used against them for surveillance, for automating for drone, for militants to be targeted without kind of human input.

12:26U.S. officials say drone strikes, rather, have killed two senior ISIS leaders. Also, how they might be detected online through various AI tools. But also, it does mean that they are trying to absorb that information and learn the best way for them to go forward and also use AI. So are these just individual supporters who have experimented with AI or has this been endorsed from higher up? So far, we haven't seen groups like IS and Al-Qaeda officially talk about AI, encourage it or discourage it. But the closest we got to that was publication by ISKP. ISKP is the Islamic State Khorasan province, which is based in Afghanistan.

13:09And this is by a media group that is affiliated with ISKP. It's called Aza 'am. And Aza 'am has a flagship publication called Voice of Khorasan. The June 2025 edition prominently had an 11-page article that was about AI. And the key slogan of that article was learning about AI today is fardu 'ayn. Now, fardu 'ayn is a religious phrase which is highly significant. It's only used in grave circumstances and matters. So in the case of jihad, so jihadists would say jihad today is fardu 'ayn if there's an ingression. So it means an individual obligation on every able Muslim. So when there's an attack, Muslims are obliged.

13:54It becomes a duty, basically. It's a duty. And if you do not fulfill that duty, you will be violating God's orders. So it's a religious duty. So for those who are not versed in this scripture, does it mean something like a fatwa, something heavily loaded with religious content? I think along those lines, yes, it could be seen as a fatwa. I mean, of course, it depends on who's giving that fatwa. it is at the end of the day, a publication. Even if it's a heavyweight publication, it's a publication. And some people might consider it, oh, it's that publication or the editor or the editorial team's view.

14:27But it was passed on as this is a fatwa, it's an obligation. So it came up with that phrase that learning about AI is not just a good to know or an optional. It is an obligation today, a religious obligation. that publication a few days later went back and removed that bit of individual obligation because there was a backlash some rivals of is mocked the group and they said we can't believe they used such a religious important phrase and applied it to ai but it gave us the impression or the idea that they actually are considering it seriously they do think that learning about ai and adopting AI tools is significant at this point.

15:13As Mina said, neither Al-Qaeda or Islamic State has come close to actually coming out explicitly and saying you must use AI. But we've seen this year there was a quite significant contribution from Mayas and Naber about media jihad and that people should use all the tech available to them. You are listening to The Global Jigsaw from BBC Monitoring. so for now they're using ai as a learning tool and as a tool to translate their material and transcribe and a few experimentations with braver stuff in terms of the translation because that would give them an outlet to a bigger audience how well is it doing you would have thought it would be the area where they would really embrace it because they really want to spread their message across various platforms and reach that global audience.

16:06We've seen media groups use AI translations. It hasn't been systematic. And our suspicion is that because they are concerned, if you don't speak the target language, I won't be able to check whether it includes any hallucinations or if there's any accuracy in the IS message or if a Quranic verse or any religious scripture has been mistranslated or twisted, I mean, that would be catastrophic, really. And it would backfire. So we hear about AI hallucinating. That's when it gets it wrong, right? Yes, essentially, when it actually makes up information that's either not in the original source, or that it's generated itself.

16:48I think that's probably more complex than that. But that's the general thrust of hallucinations on AI. Imagine if they encourage people to take that to Arabic publication or propaganda and just to use AI to translate, you would get so many different versions with really mistranslations, with errors that the people who are using it can't really detect. So I think that could be one of the reasons holding them back from aggressively using machine translation and running with it. Jihadist misgivings about the accuracy of machine translation are not misplaced. Media organisations have similar concerns, emphasising the importance of human oversight.

17:29There's also a question of authority. I don't think they'd want to give a blanket encouragement to AI. The last thing they'd want is support us actually going on to chat GPT to get fatwas, for instance. I mean, they're quite alarmed as it is by people seeking advice from what they call internet clerics, who don't have this sort of accreditation from the jihadist groups. The idea that someone's actually going to one of the Czech GPT or co-pilot and ask, what's the factor on this particular subject? I think they'd pull their hair out if they thought that was happening. America is changing. And so is the world.

18:04But what's happening in America isn't just a cause of global upheaval. It's also a symptom of disruption that's happening everywhere. I'm Asma Khalid in Washington, D.C. I'm Tristan Redman in London. And this is The Global Story. Every weekday, we'll bring you a story from this intersection, where the world and America meet. Listen on bbc.com or wherever you get your podcasts.

18:32This is the documentary from the BBC World Service. Let's explore an individual account that has stood out with its striking AI-made material, with a Russian-sounding username, Ruslan. Ruslan is an IS supporter who emerged last year, and we saw that account's activity mostly on Facebook, and it was also on Telegram. And he was a bit, or he or they, she, I don't know, kind of a bit outside the fold of the traditional IS supporters. That account was content creator, content producer, and trying to offer his services to others and saying, look, I created this slick video using AI, and these are the tools.

19:19And he would always end his video and say, this is how I created it. Some seemingly young supporters were really excited, and they would ask, oh, brother, tell us, how did you create this? That's amazing. And he'd always say, this is dead easy. I didn't in a few minutes. And the productions that Ruslan made, they were very gimmicky, very game-like, very juvenile and amateurish. However, they might appeal to young audiences, to young people who are interested in gaming because it did really look like kind of the gaming animation. He had a range of the images, kind of like postcards from around the world, kind of like I Wish You Were Here, Eiffel Tower, London Bridge, Statue of Liberty with an image of like an animated representation of an IS militant or army of IS militants in Rome, in Paris, in London, and of course some slogans.

20:11He used that also, he used it with a Neba editorial, IS's weekly paper, and he provided AI narration, paired it with all these kind of gimmicky montages to create something that, again, is more engaging, is more appealing to a young audience. That disappeared this year, the Ruslan. Why would that be, you think? We can never be sure. Sometimes they might come up in a new iteration, but we haven't seen any further propaganda that looks like it's the same person under a new name. We see jihadists online asking each other where so-and-so gone and they don't know. And sometimes they would say that so-and-so has been arrested and there's been recently warnings about online security and even more paranoia.

20:51Do you think there is a little bit of realisation that those gory videos of the past, the days of the caliphate, this realisation that doesn't really win hearts and minds and they need to create something that's more appealing. I don't think so. I think that they've got a hardcore constituency and I don't think it's just the fact that they're targeting their propaganda at a hardcore constituency. That's the audience they want. I think by their own logic, any Muslim who was squeamish about them killing infidels according to God's plan, which is what they think they're doing, is not the sort of supporter they want.

21:24Anyone who's squeamish about beheading videos, it might not be nice to watch, but this is a crucial part of our propaganda. They're very uncompromising in terms of hearts and minds. Al-Qaeda are relatively more flexible. For instance, Al-Qaeda don't do execution videos or very, very rare examples. They tend not to have gory images of people they've killed in attacks. They might issue pictures of a few corpses, but that's about as far as they would go. Now, IS, I think it enjoys its gory propaganda. We had a video from IS just a few days ago, and it was from its so-called Central Africa Province branch that is active in Democratic Republic of Congo.

22:04It was full of violence and gore. That hasn't gone away and definitely there's fewer videos just because IS does not have the media and production capacity it used to have before. In terms of have these AI animation videos replaced them, it could be they're catering for a different audience, but also it could be because of the limitations of producing ultra-violent footage and videos using AI tools. For example, ChatGPT and Copilot, if you run IS's weekly paper on NEBA, and if there's incitement in that publication, that's just text, it will refuse to translate it. And the same, if you ask it for information about bomb-making or targets, it will refuse to help with that.

22:52So there are guardrails within these AI tools that maybe prevent bad actors or supporters to use it to produce violent footage. I'm sure they could find ways around it, but we haven't seen that for now. Using machine intelligence in jihadist propaganda isn't straightforward, not just because of their own concerns about the tech, but also because of the built-in guardrails on many platforms. So does this hold them back from using AI to spread their message? I think they're excited. They're watching and learning at this point, experimenting a bit. It's very likely that they will exploit AI. The focus is about how AI is being weaponized.

23:38And that seems to be the interest. They're not focused on the media use. They're talking about how it's being used in drones, how it's being used in surveillance. It seems that they are keen to learn how they could also leverage AI to use it in the future. But if they do, they're not going to be shouting and screaming about it now. They'll be quietly using it and enhancing their use until fully confident. It would seem that using AI for propaganda purposes, as attractive as this may be, cannot be done at the expense of their values. The ISKP magazine, it talked about responsible use of AI, even had scripture from the Koran talking about not being excessive in your use of anything.

24:23And they tried to apply that to technology. And they said, we have to be very careful. And believe it or not, they talked about we have to be responsible and avoid any instances of disinformation and misinformation. and they said in their words, we cannot be like the infidels who are using AI to dupe audiences and do these misinformation campaigns. We have to seek the truth. We have to use it in a righteous way. So that was their message. And they were telling their audiences, learn about it, but be cautious and don't run with it and use it responsibly. There's real risk for jihadists if the tech backfires.

25:00It could damage their image and reputation and they know it. I think the big concern is the potential of AI and deep fakes to be used in misinformation campaigns. It's these situations when you have a lot of anger, a lot of tension, and then you could have actually a video drop or an image that looks authentic, but that's not. It could be someone burning a copy of the Koran. It could be someone insulting Islam or the Prophet and jihadists inciting, saying, OK, because of that, you have to stage attacks in this area or against this person or this organization because of this video. Back in August, an apparently AI-generated image appeared to show Druze fighters opening fire on a mosque in Sueda in southern Syria.

25:52When jihadists shared the image, they added fuel to the unfolding sectarian violence there. British journalist Alex Crawford was on the ground to witness it. They have stopped thousands of Arab tribal fighters who were battling Druze militia inside Sueda. There was a lot of very violent clashes. There's a lot of material which is going viral on social media, and that is incentivising these fighters and the fighting. It was debunked later. It was a fake video. We caught the jihadists red-handed that they put out this video. It turned out to be fake. because they're also worried about their reputation.

26:32They want to at least be seen as only sharing authentic material because one fake video could do a lot of damage for the group. Another example of fakery, albeit not fatal, prompted plenty of mockery. It was a blunder the Islamic State group won't forget in a hurry. It happened before the AI era, but it still echoes today. In 2015, IS was caught red-handed, no pun intended, using fake blood. In one of the videos from Yemen, they claimed that they were attacking this group or the Yemeni army or the Houthis. And the alleged victims had blood, but it turned out to be fake blood. It was actually Al-Qaeda that called them out.

27:17And they said one of the defectors from IS was actually part of the acting crew and told us it was Vimto, which is a very popular drink that's very dark red. It did so much damage for IS. For months, IS was ridiculed. It was a mockery because of that. And they called it the Vimto Caliphate for a long time. So that was just one instance that really undermined the credibility and image of the group to their supporters, of course, to their followers. So in terms of the audience, the target audience, you mentioned that they want to attract young people. You mentioned Facebook catching them using AI and then Instagram.

27:59And I'm curious because I remember in the old days when jihadists were on big platforms like Twitter and Facebook, and they were chased out of those platform and kind of pushed into these encrypted messaging apps. How do they get those young supporters? That's the dilemma for online jihadists. You want a secure platform that's robust and has some longevity, but they also want reach. And as you said, the big platforms always going to be subject to crackdowns. The Digital Services Act will, for the first time, spell out how big tech should keep users safe. This also includes hate speech and terrorist propaganda.