Unraveling the Threat of Deepfakes

BY BILL HOWARD – MANAGER OF IT SERVICES

Discussion regarding the many facets of AI (Artificial Intelligence) is spreading around the world. Several fears are being expressed about AI in general, but one serious threat that has come to fruition is the Deepfake. According to Co-Pilot, Microsoft’s AI assistant, “A deepfake is synthetic media that has been digitally manipulated to convincingly replace one person’s likeness with that of another.” Deepfakes include images, videos, or audio recordings that misrepresent someone as doing or saying something they never actually did or said. Several celebrities and political figures have fallen victim to these forms of deepfake. Many deepfake videos circulate throughout social media platforms causing unknown numbers of users to be coaxed into believing that their favorite celebrity is giving away unbelievable amounts of money or gifts. None of the gifts are real, but the goal was to convince you to submit data about yourself, leaving you vulnerable to a future attack or theft of your identity.

Deepfakes are created using AI along with data such as audio, video, or photographs of real people. The more data available to AI for creation, the more convincing the deepfake can be. Therefore, the best way to protect yourself from being deep faked is to protect and limit the amount of personal data available to others. Items such as high-quality images and videos of yourself should not be shared publicly. You should be careful to limit phone conversations with unknown callers as your voice can be recorded and manipulated to make it sound as if you said something you did not.

Research is currently underway to develop tools to detect and limit the use of deepfakes, but users need to remain vigilant when communicating via electronic formats as deepfakes can be inserted into virtual meetings or phone conversations. A recent real-world example was a video conference among corporate board members where only one attendee was a real person. All other attendees were AI generated deepfakes. The deepfakes were able to convince the real person to act upon a fraudulent deal costing the corporation dearly.

There are many informative uses for AI, and I do encourage everyone to explore its many facets. Please keep an understanding of care for your own personal data. To start your own exploration, try using Microsoft’s Co-Pilot, Google Bard or ChatGPT.