The “deepfake” phenomenon — using machine learning to generate synthetic video, audio and text content — is an ominous example of how quickly new technologies can be diverted from their original purposes. Month by month, it is becoming easier and cheaper to create fakes that are increasingly difficult to distinguish from genuine artefacts.
At a workshop in September 2019, IRGC convened an interdisciplinary group of experts to assess the potential risks that deepfakes might pose: from personal abuse and reputational damage to the fabrication of evidence and the manipulation of public opinion.
Later this year, we will publish a series of recommendations drawing on the workshop proceedings.
- Forging authenticity: governing deepfake risks, 2019, Policy brief (forthcoming)