As clear as the potential of deep abuse perhaps, there are still more weird new examples appearing. The Philadelpha Inquirer and Harrisburg Patriot-News report that police have arrested Raffaela Spone, a resident of Chalfont, Pa., for allegedly using deepfakes in an attempt to oust rivals from her daughter’s cheerleader team, the Victory Vipers. According to law enforcement, Spone sent the coaches AI-edited photos and videos of the teens to depict them drinking, smoking, or nude.
Police were first notified in July when one of the victims received messages from an anonymous number, but this has led others to tell similar stories. Agents linked the messages to Spone by tracing the numbers on a telemarketer-oriented site and then identifying them to an IP address used at Spone’s home. A search of the woman’s smartphone found evidence linking her to the numbers.
Spone faces criminal charges of cyber-harassing a child and related offenses. Police did not take action against the girl as there is no evidence that she was aware of her mother’s actions.
Social networks like Facebook, TIC Tac and Twitter have already banned deepfakes, making them less likely to spread. That doesn’t preclude direct harassment campaigns, however, and the incident underscores how easy it is to produce fakes – you don’t have to be an expert to create plausible images. Although there is research on spot deepfakes, detection systems like these will only be useful if people both suspect fakes and catch them before there is significant damage.