3 days Ago By Sharveya Parasnis
Explainer Briefly Slides A major controversy has erupted in Maharashtra over deepfake audio clips publicised by the BJP, allegedly featuring opposition politicians and officials discussing financial matters. “Gaurav, we need to ensure 50 crore by next week. It must be delivered to somebody in Dubai.
Please ask your friend to have cash ready...
” Amitabh Gupta, Police Commissioner, in a conversation with Gaurav Mehta, an employee of audit firm Sarathi Associates, reveals a deeper...
pic.twitter.com/T3s1ONWLW3 The audio clips were first revealed in a press conference by National Spokesperson Sudhanshu Trivedi, on the eve of the election on November 19, containing conversations purported to be between Maharashtra Congress Chief Nana Patole , NCP (SP) MP Supriya Sule , IPS Officer Amitabh Gupta and Gaurav Mehta, an employee of audit firm Sarathi Associates.
The BJP also posted these clips on their official Twitter account. They alleged that the conversations revealed illicit financial practices and cryptocurrency use by the Maha Vikas Aghadi (MVA). However, multiple news organisations and fact-checkers found the audio clips AI-generated or manipulated.
Supriya Sule rejected this claim and instead filed a complaint with the police and the Election Commission against ex-IPS officer Ravindranath Patil, who had also made the same allegations, and Gaurav Mehta. She also declared her intention to sue Trivedi for defamation. The Enforcement Directorate subsequently raided Mehta’s home office in connection with the case.
On the other hand, Maharashtra deputy CM Ajit Pawar doubled down on the claims, stating that he recognised Patole and Sule’s voices. Hours after BJP leader Vinod Tawde distributed cash for votes in Thane, the controversy started. Deepfakes in Elections The use of a purportedly deepfake clip during an election, however, highlights the potential of the technology to cause chaos.
At MediaNama’s Deepfakes and Democracy event this year, Gautham Koorma, a researcher at UC Berkley, highlighted the difficulties of detecting deepfakes. He pointed out that while watermarking AI-generated media was a possible solution, researchers had managed to break all known watermarking techniques. “I think we should be using watermarking, but keeping in mind its downside, that it can be easily broken by a sophisticated adversary.
Sometimes it’s not even a very sophisticated adversary,” he said. He also pointed out the difficulties of detecting deepfake audio on social media, as platforms often transcoded or changed the properties of uploaded audio. “And once that happens, we see that the accuracy of detection drops a lot, sometimes higher than 10 percent.
So, it’s really hard to detect these things once they’ve gone on social media,” he said. Fact-checkers invited for a discussion after the general election of 2024 stated that they saw only a few deepfakes with limited impact, but which led to an erosion of trust. Kritika Goel, Head of Editorial Operations (India) at Logically Facts, stated that it was now possible for people to claim a real audio to be a deepfake, giving them “plausible deniability” or the “liars dividend.
” Also Read:.
