A sexual deepfake that never existed sparks a fight among minors in Malaysia: the rumor is already the damage

By Free Malaysia Today · July 15, 2026.

The facts are sparse but telling. In Tawau, Sabah state (Malaysia), eight male students aged 13 and 14 were detained after a mass brawl in a commercial area, recorded in a 27-second video that went viral over WhatsApp and in which a man is seen using a helmet as a weapon. The origin of the fight, according to the police investigation reported by Sinar Harian, was a dispute over the alleged circulation of sexually explicit images and videos edited with artificial intelligence, which reportedly used the faces of two people. The minors remain held until July 17 under article 148 of the Malaysian penal code, for rioting with a weapon. The detail that changes the meaning of the whole story comes at the end: after investigating, the police did not find that AI-generated content on any social network. The material that lit the fuse, as far as is known, did not exist.

This is a textbook case of how moral panic works in the era of generative AI: the deepfake does not need to exist to cause real harm. It is enough for the possibility to be credible. The technology to fabricate fake sexual images or videos of a real person is today accessible, cheap and fast; that alone has been enough for the mere accusation of its existence to act as a social trigger, especially among teenagers who already live much of their social life and reputation on platforms like WhatsApp. The harm does not require the file: it requires only the rumor that the file might exist.

This matters because it shifts the governance problem. The public debate about sexual deepfakes rightly focuses on detection, labeling and the prosecution of those who fabricate and distribute them. But this case shows a different and harder-to-address vector: disinformation about the existence of synthetic content, which circulates by rumor among minors and leads to physical violence before any authority can verify anything. Neither platform filters nor laws against AI-generated non-consensual pornography —increasingly common in various countries— act on a rumor without a file. The regulatory gap is not only in the content, it is in the speed and the lack of critical literacy to stop an accusation before it turns into a street fight.

More broadly, the exposure of minors to synthetic sexual content —whether real or merely feared— has become one of the harshest and least glamorous fronts of the generative AI revolution, far removed from the headlines about frontier models or arms races between labs. It is the most everyday and most damaging face of the problem: schools without clear protocols, minors without tools to handle a viral accusation, and legal systems designed for a world where material evidence still mattered more than suspicion shared in a WhatsApp group.

Our reading is that this incident fits a trend we have been pointing out: the most urgent short-term risks of AI are not the science-fiction scenarios, but the erosion of everyday social trust —what is real, who tells the truth, what evidence is enough— and that strikes the most vulnerable first and hardest, in this case thirteen- and fourteen-year-old teenagers. None of this contradicts the underlying thesis that AI, well governed, can lead us to a society of unprecedented health and abundance; but that long-term promise demands, with equal urgency, investing now in critical education, school protocols and agile legal frameworks against synthetic disinformation, so that the transition does not claim, as in Tawau, the childhood of those least to blame for the problem.