A project of the Integrity InstituteDeepfakes are AI-generated, manipulated or synthetic media that falsely depict events, often with the intention of swaying public perception or causing emotional harm.
Deepfakes are most common in political and sexual contexts. In political cases, deepfakes aim to tarnish the reputation of a public figure or attribute false ideas or inappropriate behavior to them. Rapid dissemination is crucial for political impact, since once a deepfake is unmasked, it is fairly straightforward to block its further spread. Because they are typically not distributed through fact-checked channels, social networks play the primary role for the distribution (and impact) of political deepfakes. In sexual cases, abusers create fake videos in order to harm the target's reputation, for blackmail purposes, or for personal/sexual use. The violation of the target's right to control their image and likeness is a key concern in all instances, though reputation and financial harms can also be devastating.
Platforms play a role in facilitating deepfake creation by hosting data such as a user's voice, appearance, and mannerisms, which are essential for generating believable deepfakes. Moreover, platforms provide tools for uploading, editing, and disseminating content, enabling the widespread distribution of deepfakes. The prioritization of engagement by these platforms amplifies the impact of deepfakes, as their sensational nature thrives in environments of maximum user engagement, leading to increased distribution and potential harm.