AI is increasingly being used as a cover by politicians when embarrassing audio and video clips of them emerge.

This report uncovered the first high-profile case of “liar’s dividend,” the ability of the powerful to claim plausible deniability of unflattering footage. When embarrassing audio footage of an Indian politician accusing his own party of corruption emerged, he dismissed them as AI-crafted deepfakes. We conducted three independent forensic evaluations on the audio clips and determined one might be partially fabricated, while the other was undeniably genuine.

The reporting drew international attention to liar's dividend and was cited in multiple congressional testimonies, including one before the U.S. Senate hearing focused on the need for transparency in AI. The piece was citied in a TED talk in December 2023. (The reporting followed my 2020 news break on the first use of deepfake in an Indian election.)

Read full story