For much of modern history, the camera was regarded as an impartial witness. A photograph or video recording carried the weight of evidence. It could settle disputes, support investigations, and shape public opinion because people generally believed that seeing was believing. That certainty is now disappearing, with consequences that extend far beyond technological change. It is reshaping justice, influencing elections, affecting conflicts, and challenging the public’s ability to distinguish fact from fiction.
The threat is no longer limited to fabricated images and videos. Increasingly, authentic footage is being dismissed as fake. This phenomenon, widely known as the “liar’s dividend,” allows powerful individuals to deny genuine evidence of misconduct simply by claiming it was generated through artificial intelligence. Long before AI-generated videos became widely accessible, governments, political figures, and institutions accused of wrongdoing often questioned the authenticity of damaging evidence. Today, those denials carry greater weight because public trust in digital content has been steadily eroded. In an environment where almost anything can be manipulated, even the most compelling visual evidence can be cast into doubt, making truth itself a matter of public dispute rather than established fact.
A History of Manufactured Doubt
Deepfakes did not create the habit of dismissing inconvenient evidence. They have simply made it far more convincing. This has transformed a long-standing tactic into a powerful tool for political manipulation, conflict, and public deception.
In the United States, AI-generated images showing Donald Trump being arrested spread across social media in March 2023 after reports emerged that he could face indictment. The images were viewed millions of times before many users realised that no arrest had taken place. The episode demonstrated how fabricated content could shape public perception long before the truth caught up.
In Slovakia, the technology moved beyond misinformation to potential election interference. Just two days before the 2023 parliamentary election, during a legally enforced media blackout, a fake audio recording appeared to capture opposition leader Michal Šimečka discussing election rigging with a journalist. Although the recording was entirely fabricated, the reporting restrictions prevented immediate public debunking, allowing the audio to circulate unchecked before voters cast their ballots.
The Russia-Ukraine war revealed another dimension of the threat. Weeks after the 2022 invasion, hackers broadcast a deepfake of President Volodymyr Zelensky apparently urging Ukrainian troops to surrender. Although the video was quickly exposed, it demonstrated that synthetic media could be deployed as a weapon of war to create confusion and undermine public confidence.
In India, Prime Minister Narendra Modi also became the target of AI-generated content when a fabricated video showed him dancing to a Bollywood song during the 2024 general election campaign. The video attracted tens of millions of views, prompting Modi to warn that deepfakes had become a serious threat in a country where many people struggle to verify the authenticity of online content.
Pakistan itself offers one of the starkest examples, and a genuinely strange one, because the same politician ended up on both sides of the technology. Ahead of the February 2024 general election, jailed former prime minister Imran Khan’s party used an AI voice clone, openly disclosed as synthetic, to let him address campaign rallies and eventually claim victory from behind bars, a deliberate, consented-to use of the technology to get around a ban on public campaigning. But barely a day later, Khan became a victim of the very same technology. On the eve of polling day, a deepfake audio clip falsely attributed to him began circulating on social media, claiming he was calling on PTI supporters to boycott the vote entirely. PTI had to scramble to publicly denounce it as fabricated, and independent fact-checkers confirmed it was fake.
Pakistan offers one of the clearest examples of deepfakes reshaping politics. In a striking twist, the same politician found himself on both sides of the technology. Ahead of the February 2024 general election, jailed former prime minister Imran Khan’s party used an AI-generated voice, openly identified as synthetic, to deliver campaign messages and later claim victory from prison. It was a transparent use of AI to overcome restrictions on public campaigning. Within a day, however, Khan became a victim of the same technology. On the eve of polling, a fake audio recording circulated on social media claiming he had urged PTI supporters to boycott the election. The party quickly dismissed the clip as fabricated, and independent fact-checkers confirmed it was fake.
For women in Pakistani politics, the consequences have been even more damaging. Punjab Information Minister Azma Bukhari said she was “shattered” after a sexualised deepfake video placed her face onto another woman’s body in an apparent attempt to discredit her. Punjab Chief Minister Maryam Nawaz Sharif has also been repeatedly targeted with fabricated videos and doctored images. Pakistan’s Federal Investigation Agency has arrested suspects accused of circulating manipulated images of her through coordinated social media accounts.
Regional lawmaker Meena Majeed was similarly targeted with a fake video showing her embracing a male chief minister. The clip was presented as an attack on her province’s honour. As Pakistani media researchers have noted, in a conservative society such fabrications do more than spread misinformation. They can destroy reputations, damage political careers, and discourage women from entering public life.
The India-Pakistan conflict is the clearest case study
The recent India-Pakistan military conflict exposed just how quickly AI-generated misinformation can spread during a crisis. Within days of the hostilities, fact-checkers and digital forensic experts were overwhelmed by a wave of synthetic content, including cloned voices of military spokesmen, fabricated claims about fighter jets being shot down, and manipulated videos designed to inflame public opinion before they could be verified.
One widely shared video falsely showed a Pakistani military spokesman admitting the loss of fighter jets. Forensic analysis later confirmed it had been generated using AI, but only after several media outlets had repeated its claims. Elsewhere, old video game footage was presented as real battlefield scenes, prompting officials on both sides to urge the public not to mistake simulation for combat.
The misinformation campaign created more than confusion. It became a security threat. While fabricated content spread online, genuine videos showing troop movements, missile launchers, and military equipment were also widely shared by civilians. Unlike deepfakes, this material provided real-time intelligence that could be exploited by the opposing side. Military authorities in both countries appealed to citizens to stop filming and posting sensitive locations, acknowledging that smartphones had become part of the modern battlefield.
The greatest damage, however, was psychological. As the line between authentic and fabricated content became increasingly blurred, public trust deteriorated. Genuine images of destruction were dismissed as fake while fabricated ones were accepted as reality. In the fog of digital war, uncertainty became as powerful as any weapon.
Why the problem resists control
One of the greatest challenges in tackling deepfakes is jurisdiction. The platforms where manipulated content spreads most rapidly are often based overseas, operate under foreign laws, and face little immediate accountability for the political consequences of viral content in Islamabad, Washington, New Delhi, or Bratislava. Governments can ban an app or summon local representatives, but they have limited power to force global technology companies to remove harmful content within the critical hours that matter most during an election, a conflict, or a national crisis.
The Search for Global Standards
Governments and international organisations have begun responding to the rise of deepfakes, but regulation continues to trail the technology. The European Union’s AI Act requires AI-generated content, including deepfakes, to be clearly labelled, although many of its transparency provisions will only be fully implemented by 2026. The United Nations has also adopted ethical guidelines and resolutions promoting safe, transparent, and accountable AI, while UNESCO has called for stronger global standards. Independent organisations have stepped in where regulation has fallen short. Research institutions, digital forensic laboratories, and human rights groups now play a critical role in detecting manipulated media and helping journalists verify content during elections and conflicts. Despite these efforts, experts agree that the global response remains fragmented. Most safeguards exist as guidelines or voluntary commitments rather than enforceable laws, leaving governments struggling to keep pace with a rapidly evolving information landscape.
Pakistan has been slow to build a legal framework for the age of artificial intelligence. Although the country approved its first National AI Policy in 2025, it remains largely a roadmap rather than an enforceable regulatory framework. The primary law governing online content is still the Prevention of Electronic Crimes Act (PECA) 2016, which predates generative AI and contains no specific provisions for deepfakes, voice cloning, or synthetic media. A 2025 amendment to PECA introduced an offence for spreading false or fake information, but rights organisations have argued that its broad wording could be used against journalists and political critics rather than those creating AI-generated deception. Despite thousands of AI-related complaints reaching the Federal Investigation Agency, Pakistani law still provides no legal definition of a deepfake. As a result, Pakistan continues to respond to deepfakes on a case-by-case basis through official denials, FIA investigations, and requests for content removal. Digital rights experts have called for targeted legislation that clearly defines synthetic media offences while strengthening data protection and independent oversight instead of relying on broad censorship powers.
Fighting Back Against Deepfakes
The challenge is not impossible to overcome, but it demands more than symbolic action. News organisations need stronger verification teams capable of debunking viral falsehoods within hours rather than days. Governments must work together to hold technology platforms accountable, requiring them to label AI generated content and respond rapidly during elections and national emergencies. Media literacy should become a core part of education, equipping citizens to question digital content before sharing it. At the same time, armed forces must provide clear guidance on what civilians should and should not post online, because in modern conflict, an ordinary smartphone can reveal information that puts lives at risk.
The camera was never completely neutral. People have always decided where to point it and what story to tell. Yet it once carried the authority of a witness. Restoring that trust will require more than new laws or better detection tools. It will depend on governments, technology companies, journalists, and citizens working together to protect the integrity of information. In an era when deception can be created with a few clicks, defending the truth is no longer just the responsibility of the media. It is a responsibility shared by society as a whole.
