Fighting sharply escalated along the Pakistan–Afghanistan border after cross-border attacks and airstrikes.
Pakistan launched air and ground strikes against Afghan Taliban positions, declaring an ‘open war,’ while Afghanistan says it is retaliating for Pakistan’s violations of its sovereignty. Both sides accuse each other of aggression, ending a fragile ceasefire and raising fears of wider conflict.
In the above context, a video of three women has gone viral on social media. In the video, the women are seen speaking in Pashto.
The video is being shared with a caption suggesting that Afghan women have appealed to the Pakistani army to free them from the Taliban’s brutal rule.
A social media user shared the video with a caption that read, “Afghan women thank the Pakistan Army for the Pakistan Air Force’s action against terrorists.”
SouthCheck found the claim to be false, the viral video is AI-generated and not authentic.
A hyperlocal Pakistani channel, Khyber TV, shared the clip on its Facebook page on February 23 with a caption implying the video was an AI creation. The caption stated that Khyber News TV had aired a fake report claiming Afghan women supported Pakistani airstrikes and demanded an end to the ‘transitional’ regime, and that the words attributed to a woman were manufactured using artificial intelligence and not real. It also noted that the report had sparked strong and sarcastic backlash on social media.
However, it’s not clear from the available information whether the women shown in the video are from Pakistan or Afghanistan, or what event the footage was initially recorded at.
Closer visual analysis reveals signs common in AI content; for example, the woman in the middle is speaking, while the two women beside her remain unnaturally still and don’t blink for long stretches. The lighting on their faces also appears inconsistent: the central figure is lit from the front, while the woman in green has light coming from the left. These anomalies suggest manipulation.
We ran the clip through an AI detection tool and found that while the video itself wasn’t flagged as AI-generated, the audio was identified as 94.4 % likely to be AI-generated. Tools like these can miss signs of manipulation, so their results aren’t definitive on their own.