After a series of scams involving dishonest job seekers who charge a signing bonus and run, some Fortune 500 companies have begun testing software that can detect a imitation real person during a live video call.
The detection technology was developed courtesy of Get Real Labs, a recent company founded by Only Faridprofessor at the University of California, Berkeley and a recognized authority in the field of deepfakes and image and video manipulation.
Get Real Labs has developed a set of tools to detect images, audio, and video generated or manipulated using artificial intelligence or manual methods. The company’s software can analyze a face during a video call and detect clues that may indicate it has been artificially generated and swapped for a real person.
“These are not hypothetical attacks, we are hearing about them more and more often,” Farid says. “In some cases, it seems like they are trying to gain intellectual property by infiltrating a company. In other cases, it seems like it is purely financial, they are just taking a bonus to sign a contract.
The FBI released warning in 2022 about imitation job hunters who assume the identity of a real person during video interviews. British design and engineering company Arup lost $25 million for a imitation scammer posing as a corporate CFO. Romance scammers have also adopted the technology, swindling unsuspecting victims out of their savings.
Impersonating a real person during a live video broadcast is just one example of the reality-sinking fraud now made possible by artificial intelligence. Huge language models can convincingly mimic a real person in an online chat, and miniature videos can be generated using tools like Sora from OpenAI. Impressive advances in artificial intelligence in recent years have made deepfakers more convincing and accessible. Free software makes it easier to hone your deepfaking skills, and readily available artificial intelligence tools can turn text prompts into realistic-looking photos and videos.
However, live video impersonation is a relatively recent field. Creating this type of deepfake typically involves using a combination of machine learning and facial tracking algorithms to seamlessly blend the imitation face with the real one, allowing the attacker to control what the illicit likeness appears to say and do on screen.
Farid gave WIRED a demonstration of Get Real Labs’ technology. When you view a photo of a company’s conference room, the software analyzes the metadata associated with the image for signs of tampering. Several major AI companies, including OpenAI, Google and Meta, are now adding digital signatures to AI-generated images, providing a resilient way to confirm their inauthenticity. However, not all tools provide such tags, and open source image generators can be configured not to do so. Metadata can also be easily manipulated.
GIF: Will Knight