Boko Haram, ISWAP Fold Generative AI Into Propaganda Push

The New York Times report lands on something counterterrorism researchers have been tracking for a while, that groups including Boko Haram are folding generative AI into how they run propaganda and recruitment. Take the piece's specifics as newly surfaced rather than settled here, because the sturdiest grounding I can offer comes from the wider trend other outlets have documented, not the NYT's own reporting.

The clearest documented case in the Lake Chad Basin comes from Africa Defense Forum, which quotes Bulama Bukarti of the Tony Blair Institute for Global Change describing how Boko Haram and the Islamic State West Africa Province are using TikTok to host live programs and answer user questions. On top of that, the Islamic State runs a program called News Harvest featuring AI-generated news anchors, and ADF reported that the group had circulated a guide in 2023 on how to use generative AI securely, followed by an al-Qaida media-wing AI workshop in February 2024. This is current tradecraft, not a hypothetical.

The stakes are the cost curve. Slick propaganda used to require studios, editors, and time. Voice cloning, deepfake video, and cheap chatbots collapse that toward zero. Daniel Siegel, a Columbia University researcher quoted by ADF, argues that embedding extremist narratives inside content that mimics the tone and style of popular entertainment lets material slip past the usual scrutiny. An interesting inversion has already shown up on the defensive side, with al-Qaida-affiliated groups warning members that adversaries could use AI-generated audio to give fake commands to followers.

The honest caveat is that this coverage does not tell you how much of a battlefield edge AI actually gives Boko Haram or ISWAP, or how many recruits it converts versus older in-person channels in the Lake Chad Basin. It also does not name the specific consumer AI products being used, or whether the underlying models are open-weight or hosted, which is the answer every vendor and regulator will eventually be asked for.

If there is a forward-looking read, it is that the interesting policy fight is moving from whether a frontier lab can block bioweapon uplift to whether consumer platforms and mid-tier AI vendors can keep a live insurgency out of their recruitment loop. That is a very different capability question, and the constituencies pressing on it, in Abuja and Washington and any platform with reach in West Africa, are about to get louder.