Former Prime Minister of Pakistan
Imran Khan
, currently in jail on charges his supporters say are politically motivated, appeared to deliver a
victory speech
Saturday night after his party won the most seats in the country's general election. But Khan did not give the speech - an
artificial intelligence
mimic of his voice did.
According to The New York Times, Khan's
Pakistan
Tehreek-e-Insaf (PTI) party created the video using
AI
to simulate Khan's voice based on notes he passed from jail. The video shows images of Khan and his supporters overlaid with audio of his voice declaring victory and alleging election fraud by opponents while displaying a disclaimer created with AI.
Chairman Imran Khan's victory speech (AI version) after an unprecedented fightback from the nation
Experts said this demonstrates both the potential for using AI to circumvent suppression of speech and the technology's capacity to mislead. Khan has rallied supporters throughout the election while jailed, using online platforms and now AI to get his message out despite efforts by Pakistan's military leaders to censor him.
"In this case, it's for a good end, perhaps an end we'd support - someone who's locked up on trumped-up charges of corruption being able to speak to his supporters," said AI expert Toby Walsh, as quoted in The New York Times. "But at the same time, it's undermining our belief in the things we see and hear."
The AI video capped a tech-centric election strategy for Khan after he was barred from running himself. His party held online rallies and used social media despite censorship.
The use of AI to replicate Khan's voice builds on examples like South Korean President Yoon Suk Yeol's campaign creating an AI avatar for campaigning purposes, as stated in The New York Times. Last week, US President Biden's voice was mimicked in an apparent voter suppression robocall in New Hampshire using AI, after which the FCC banned the AI-generated voice in robocalls.
The use of AI technology has raised some concerns regarding its credibility. It is now possible to create fake videos using AI that could potentially show politicians saying things that are inflammatory or making false declarations. The concern is that viewers may not realise that the videos are fake, even if they are labelled as AI-generated.
English (US) ·