Hyderabad: Social media users are questioning a viral video of slain Liberation Tigers of Tamil Eelam (LTTE ) chief Velu Pillai Prabhakaran in which he is vowing revenge for allegedly suppressing the rights of Tamils by the Sri Lankan government.
In the video, Prabhakaran could be heard saying, “My Dearest and Respected People of Tamil Eelam,we were compelled to wage war with weapons because the oppressive Sinhalese government denied us our rights and unleashed severe repression on our people. We never sought war, but we stood firm in defence of our rights. Despite our willingness to engage in peace talks on multiple occasions, our struggles were brutally suppressed in 2009 with the aid of global imperialist forces, led by the Mahinda and Sinhala government, setting us back 15 years during the ceasefire period.”
“Our struggle has faced numerous betrayals, yet we have always risen above them. Today, the betrayers who stand before you must be defeated, and the sacrifices of our heroes and common people must be honoured. Unite as Tamils and defeat these forces. The ultimate aim of the Tigers is the motherhood of Tamil Eelam,” he adds.
Fact Check
Newsmeter found the video to be generated with Artificial Intelligence (AI).
V Prabhakaran, the leader of LTTE was killed in the war in 2009 with the Srilankan army. But, several conspiracy theories are still making the rounds, with few Tamil nationalist organizations and leaders claiming that he is still alive.
Upon close examination of the video, it becomes apparent that only Prabhakaran's head is moving, while the rest of his body remains static—strongly suggesting that the video is AI-generated.
The Deepfakes Analysis Unit (DAU) of the Misinformation Combat Alliance (MCA), an Indian cross-industry collaborative that includes NewsMeter as a member, analyzed the video.
They used the deepfake detection tool Loccus.ai to evaluate the audio track. This tool assigns a confidence score to determine the authenticity of the audio, with scores close to 1.0 indicating real audio and scores near 0.0 suggesting AI generation. In this case, the analysis produced a score of 0.253, suggesting that the audio was likely AI-generated.
The second analysis was done with TrueMedia, which analyses the face and audio in the video individually. The tool adds a confidence score to both the audio and face separately. As per the result, it can be concluded that both the face and audio are tampered with AI.
Further, analysts also examined the video using Hive AI, a tool designed to assess the authenticity of both the audio and visuals. Hive employs separate tools for each analysis.
Hive's visual detection tool identified multiple instances of manipulation throughout the video. Its audio analysis tool segmented the track into 10-second clips, analysing each segment individually. The tool flagged the majority of the audio clips as likely tampered with by AI, although it determined that a few segments were 'Not AI.'
Given the results from all these tools—Loccus.ai, TrueMedia, and Hive—it is clear that the video was artificially generated and is not an original recording.