Sarah Barrington

The proliferation of AI-generated audio (so-called "deepfake" audio) poses significant threats to information integrity, from voice cloning fraud to synthetic music copyright disputes. We present a temporal coherence analysis framework built upon Contrastive Language-Audio Pretraining (CLAP) embeddings that spans speech, instrumental music, and music with vocals.

Perspective Sarah Barrington, Hany Farid / Mar 31, 2026 The Sora 2 introduction is displayed on a mobile phone with the OpenAI company's branding seen in the background, in this photo illustration in Brussels, Belgium, on October 19, 2025. (Photo by Jonathan Raa/NurPhoto via AP) When it first launched in November of 2025, OpenAI’s Sora was in high demand, thanks in part to the clever invitation-only roll out and the ground-breaking technology that created highly realistic videos.

This perspective is part of a series of provocations published on Tech Policy Press in advance of a symposium at the University of Pittsburgh's Communication Technology Research Lab (CTRL) on threats to knowledge and US democracy. Last month, OpenAI unleashed Sora 2, the latest in a surge of generative-AI technologies empowering users to create perceptually realistic media from a single image or prompt.