Tech firms combat deepfake scams fueled by AI

Tech companies are working hard to fight against deepfake scams, which use advanced artificial intelligence (AI) to create fake voices and videos. Criminals are increasingly using these technologies to trick people into giving away money or personal information. One story highlights the experience of Debby Bodkin. Her 93-year-old mother received a call from what seemed to be a cloned voice of her grandchild saying, “It’s me, mom... I’ve had an accident.” Thankfully, the granddaughter answered the call and realized it was a scam. Bodkin noted that these types of calls happen daily. Deepfake scams typically involve scammers convincing victims to pay for fake medical treatments or other false emergencies. Criminal groups have even used AI-generated avatars of company employees to swindle money. In Hong Kong, a worker was scammed out of HK$200 million (around US$26 million) after attending a video call with these fake avatars. Research shows that only about 0.1% of people in the U.S. and the U.K. can recognize a deepfake accurately. This is a stark contrast to earlier times when it took much longer to create convincing voice replicas. Now, it only takes seconds to reproduce a human voice, making it easier for scammers. Companies like Intel are creating tools to identify AI-generated content. For instance, Intel’s “FakeCatcher” technology can spot fake images by analyzing changes in facial blood vessels. As deepfakes become more sophisticated, it’s becoming harder for the average person to detect them. Experts warn that deepfakes pose a significant global cybersecurity threat. They can damage a company's reputation or infiltrate businesses more easily, especially with more people working from home. As the problem grows, consumers are also looking for ways to protect themselves. Recently, a Chinese company unveiled a smartphone featuring a built-in deepfake detector. A British startup has introduced a web browser that flags synthetic media for businesses. Experts believe deepfakes may become as common as spam, with detection technology eventually improving to help manage the situation better. However, we are not there yet.