INDIA — According to the World Economic Forum's 2024 Global Risk Report, India ranked number one globally for the risk of misinformation and disinformation. This is a remarkable achievement for a country that also ranks first in the world for the consumption of mobile data, the production of engineers, and the asking of "what is your salary?" at family functions. We are a nation of superlatives. This particular superlative is the one we did not campaign for.

During Operation Sindoor week, BOOM fact-checkers extracted audio from viral videos of PM Modi, Home Minister Amit Shah, and External Affairs Minister S. Jaishankar and ran them through AI audio detection tools at Deepfake-o-meter, an open-source resource from the University at Buffalo. The tools confirmed: the politicians had not said what the videos showed them saying. The videos were AI-generated. The process of confirming this took a few hours. The videos had been shared four million times before the confirmation was published.

The deepfakes went in both directions. A deepfake of Shehbaz Sharif admitting Pakistan's defeat went viral in India. A deepfake of Modi apologising to Pakistan went viral in Pakistan. Both were false. Both were shared extensively by people who found them believable, which is the most revealing detail of all — not that the deepfakes existed, but that both countries' citizens found "the enemy's leader has admitted defeat" credible enough to share without checking. We will believe anything that confirms what we already think. The technology is new. The impulse is not.