Claim: PM Shehbaz Sharif has admitted defeat to India, saying Pakistan faced a “lack of resources” and “political isolation”. He laments no support from friendly or powerful countries, apart from Türkiye, and expresses fear that “Pakistan may end up being taken over by the enemy”.
Fact: The video is doctored, likely through AI. It uses authentic footage of Sharif’s speech in the National Assembly from 7 May 2025 with manipulated audio and facial expressions. The prime minister has said nothing of the sort and definitely did not concede defeat to India.
On 10 May 2025, X (formerly Twitter) user @KnightHindutva posted (archive) a video showing Pakistani Prime Minister Shehbaz Sharif, apparently admitting defeat to India in front of lawmakers during a speech in the National Assembly.
The accompanying caption — which includes the hashtags “#IndiaPakistanWar,” “#Ceasefire,” “#PakistanIsATerrorState,” and “#OperationSindoor” — reads as follows:
“Joker @CMShehbaz accepting defeat of pakistan. 🔥”
In the video, which is approximately 70 seconds long, Sharif can allegedly be heard saying the following:
“My dear countryfolk, today I’m going to state an extremely serious and painful truth to you. We are conceding defeat in the ongoing war against India. Our armed forces are fighting bravely but lack of resources, political isolation, and the enemy’s strength have put us in a difficult situation. I say this with sadness that apart from Türkiye, no country is standing with us… Not the Arab world, not China, not any other powerful country. Everyone has ignored our plea for help. My heart is heavy when I say that if the situation remains the same, then Pakistan may end up being taken over by the enemy. Our homeland, our autonomy, our dreams, everything is in danger. I request all of my compatriots; this is not the time to despair, this is the time to save Pakistan, this is the time to offer sacrifice. We won’t bow down, we won’t stop. If we all unite, no power can eliminate Pakistan. Pakistan will survive only if we defend it with our whole lives. May God be with us.”
Pakistan-India conflict
In one of the most intense military escalations in decades, Pakistan and India exchanged drones, fire, shelling, and missiles for four days in May 2025, resulting in casualties on both sides. The conflict raised fears of a nuclear war between the two neighbouring countries who have fought three wars over the disputed Kashmir region.
India launched Operation Sindoor on 7 May 2025, targeting multiple locations inside Pakistan and Pakistan-administered Kashmir. These attacks killed at least 31 people — including three children — and wounded 57 others, according to Pakistani authorities.
Information Minister Attaullah Tarar confirmed that the Pakistan Army “intercepted and destroyed 77 Israeli-made Harop drones” from India, a number also cited by state media outlet, the Pakistan Television Corporation (PTV). As of now, Soch Fact Check has not been able to independently verify the exact number of Indian drones that entered Pakistan or how many were shot down.
However, debris from multiple drones was found in a number of locations; Soch Fact Check visited and investigated the attacks at two crash sites each in Karachi, Rawalpindi, and Lahore. Read our on-ground report here.
In response, the Pakistan Army announced a counterattack — named “Operation Bunyan-um-Marsoos”, which is an Arabic phrase that translates to “steel wall” — against India in the wee hours of 10 May 2025.
The military’s media wing, the Inter-Services Public Relations (ISPR), said Pakistan hit “26x military targets as well as facilities that were used to target Pakistani citizens and those enterprises that were responsible for fomenting terrorism in Pakistan”. These bases included those at “Suratgarh, Sirsa, Bhuj, Naliya, Adampur, Bhatinda, Barnala, Halwara, Awantipura, Srinagar, Jammu, Udhampur, Mamun, Ambala, and Pathankot”, it added. “[The] BrahMos storage facilities at Beas and Nagrota were also destroyed.”
India’s Defence Ministry alleged that Pakistan “launched 300-400 drones across 36 Indian locations”.
In a 13 May statement, the Pakistan Army announced a total death toll of 51, including “40 civilians and 11 military service members”, and that close to 200, including 121 civilians, were wounded.
On the other hand, India announced on 11 May that 21 civilians and five soldiers have died.
On 10 May, both nations announced a ceasefire agreement, which Pakistan says was brokered by US President Trump. New Delhi maintains that the ceasefire was worked out “directly” by both sides.
Fact or Fiction?
Soch Fact Check reverse-searched keyframes from the viral video and found that it matches Sharif’s address in the National Assembly on 7 May 2025, hours after India launched an attack on Pakistan as part of Operation Sindoor.
An over-five-minute portion of Sharif’s speech was uploaded by The Associated Press, while a live stream is available here. The full, almost-20-minute address is available on the PTV YouTube channel here.
Nowhere in Sharif’s speech did he admit defeat to India nor did he lament a lack of resources and support from friendly nations. Soch Fact Check suspected that the same footage was manipulated using some type of speech-focused artificial intelligence (AI) tool.
In authentic speeches, the PM pauses between his words but, in the viral video in question, he continues his comment without any breaks.
We also observed that Sharif’s pronunciation of certain Urdu words is incorrect. For example, in the viral video, he says, “ajiz,” “sarjameen,” and “jinda;” however, the correct pronunciation is “aziz,” “sarzameen,” and “zinda.” Similarly, he utters the words “khudmukhtari,” “khawab,” and “khatrey” more literally, with a hard “kh,” which is unlike Sharif’s typical Urdu pronunciations.
This indicates that the fictitious speech — whether by converting one’s own voice to that of the selected individual or by the text-to-speech (TTS) method, whereby phrases used in the input are converted to the voice of the individual of choice — was generated by someone who is not of Pakistani origin.
Interestingly, Sharif’s hair and mouth appear to be blurred at some points and his teeth apparently disappear randomly during the speech.
Sound engineer’s analysis
Soch Fact Check also sought a comment for audio analysis from Shaur Azher, a lecturer and audio engineer at Soch Videos, our sister company, who specialises in sound design and mixing and mastering audio.
Azher — who also teaches at the Shaheed Zulfikar Ali Bhutto Institute of Science and Technology (SZABIST) — said he observed multiple anomalies, such as synthetic audio signatures, tonality mismatch, and unnatural delivery. “The audio in question is highly likely to be artificially generated and not a genuine recording of Shehbaz Sharif,” he noted.
The technical observations he provided to support his argument are as follows:
- Synthetic frequencies: The audio contains noticeable synthetic frequency artefacts, which are commonly associated with AI-generated or heavily-processed speech. These artefacts are not typical in natural human vocal recordings.
- Monotone delivery: The speech is unusually monotonic, lacking natural pitch fluctuations. Human speech — particularly that of Shehbaz Sharif in known recordings — typically features expressive modulation.
- Tonality discrepancy: Shehbaz Sharif’s voice typically has a deeper, slightly huskier, and more resonant tone. The voice in the clip does not match this natural tonality.
- Artificial delay calculations: There are signs of unnatural timing and pauses in the speech pattern. These delays suggest the use of AI voice generation tools, which often produce rhythm inconsistencies during synthesis.
DeepFake-O-Meter
The second tool we tested is DeepFake-O-Meter, developed by the University at Buffalo’s Media Forensics Lab (UB MDFL). Of the available detectors, we used three, namely DSP-FWA, WAV2LIP-STA, and LIPINC.
The detectors revealed that the probability of the audio in question being fake was 100%, 97.3%, and 99.9%, respectively. The results are available here, here, and here; however, only those who have signed up can view them.
DSP-FWA, or Dual Spatial Pyramid for Exposing Face Warp Artifacts, “is a deep learning-based method that can effectively distinguish AI-generated fake videos”, according to the DeepFake-O-Meter. It “target(s) the artefacts in affine face warping as the distinctive feature to distinguish real and fake images”.
WAV2LIP-STA, or WAV2LIP Spatial and Temporal Attention, “can identify wav2lip model generated deepfake videos”, according to the DeepFake-O-Meter. This deep learning-based method “learns the spatial and temporal artifacts from multi-frames (5) to expose wav2lip deepfakes”.
LIPINC, or Lip-syncing detection based on mouth inconsistency, is also a deep learning-based method that detects “lip-syncing deepfakes by identifying the spatial-temporal discrepancies in the mouth region of deepfake videos”, according to the DeepFake-O-Meter.
We have reached out to UB MDFL Director Dr Siwei Lyu, the principal investigator (PI) of the DeepFake-O-Meter and a professor at the university’s Department of Computer Science and Engineering, for a comment.
Virality
The post by @KnightHindutva has been viewed over 2,500 times so far.
Soch Fact Check found the doctored video posted here, here, here, here, and here on Facebook.
Conclusion: The video is doctored, likely through AI. It uses authentic footage of Sharif’s speech in the National Assembly from 7 May 2025 with manipulated audio and facial expressions. The prime minister has said nothing of the sort and definitely did not concede defeat to India.
Background image in cover photo: ShehbazSharif
To appeal against our fact-check, please send an email to appeals@sochfactcheck.com