“We should all have AI safe words we tell the people who are close to us in order to prevent malicious deepfakes of ourselves.”