Professor Siwei Lyu, University of Buffalo, started his research on detecting computer manipulated images and audio 26 years ago. Then in 2017 Artificial Intelligence (AI) came on the scene and he was both excited and concerned. The ability to produce realistic deepfake manipulated images, videos and audio exploded.

Professor Lyu told podcast host Frank Knapp that detecting deepfakes will no longer be possible by individuals and institutions. Only sophisticated technical examinations will be able to detect deepfakes.

The tools to create deepfakes are easy to obtain and even people with no computer education, including teenagers, can easily be deepfake creators.

Professor Lyu has a warning for all politicians regardless of partisan affiliation. Politicians are particularly vulnerable to deepfakes because of the amount of public video, images, and audio of them.

Professor Lyu’s computer lab has created a free online platform for deepfake detection, the Deepfake-O-Meter.

However, he believes that there should be legal requirements for AI generated deepfakes to be labeled, including both visual and audio warnings as well as watermarking of all AI generated deepfakes. Such state and federal laws would reduce, but not eliminate, the number of deepfakes being generated and hold deepfake creators legally accountable for the labeling.