How Intel Labs Utilize Artificial Intelligence to Combat Misinformation

2023-07-17

A few years ago, Deepfake technology seemed to be a novel technology that relied on powerful computing capabilities. Today, Deepfake technology is ubiquitous and has the potential to be used for misleading information, hacking attacks, and other dangerous purposes. Intel Labs has developed real-time Deepfake detection technology to address this increasingly serious problem. Ilke Demir, a senior research scientist at Intel, explains the principles of Deepfake technology, Intel's detection methods, and the ethical considerations involved in developing and implementing such tools. Deepfake refers to videos, voice, or images created using artificial intelligence (AI), in which actors or behaviors do not actually exist. Deepfake utilizes complex deep learning architectures such as generative adversarial networks, variational autoencoders, and other AI models to create highly realistic and credible content. These models can generate synthetic characters, lip-sync videos, and even perform text-to-image conversions, making it difficult to distinguish between real and fake content. Sometimes, the term "deepfake" is also used to refer to tampered real content, such as the manipulated video of former Speaker of the House Nancy Pelosi appearing drunk in 2019. Demir's team focuses on computer-generated deepfake content, which is machine-generated synthetic content. He said, "It's called deepfake because there is this complex deep learning architecture in generative AI that creates all this content." Cybercriminals and other malicious actors often abuse deepfake technology. Some use cases include political misinformation, adult content featuring celebrities or individuals without consent, market manipulation, and impersonation for economic gain. These negative impacts highlight the need for effective deepfake detection methods. Intel Labs has developed one of the world's first real-time deepfake detection platforms. Instead of looking for forgery traces, this technology focuses on detecting real signals, such as heart rate. Using a technique called photoplethysmography, the detection system analyzes changes in vein color caused by oxygen levels, which are visible on a computer and can be used to detect whether a person is real or artificially synthesized. Demir said, "We are trying to observe what is real. Heart rate is one of them. So when your heart pumps blood, the blood reaches the veins, and the color of the veins changes due to oxygen levels. This change is not visible to our eyes; I cannot tell your heart rate just by looking at this video. But this color change is visible on a computer." Intel's deepfake detection technology is being applied in various fields and platforms, including social media tools, news organizations, broadcasting companies, content creation tools, startups, and non-profit organizations. By integrating this technology into workflows, these organizations can better identify and mitigate the spread of deepfakes and misinformation. Although deepfake technology has the potential for abuse, it also has legitimate applications. One early application is to better represent individuals in a digital environment by creating virtual personas. Demir mentioned a specific use case called "MyFace, MyChoice," which uses deepfake technology to enhance privacy protection on online platforms. In simple terms, this approach allows individuals to control their appearance in online photos and replace their faces with "quantifiably different deepfakes" if they want to avoid recognition. These controls provide more privacy and control over personal identity, helping to counter automatic facial recognition algorithms. Ensuring the ethical development and implementation of artificial intelligence technology is crucial. Intel's Trusted Media team collaborates with anthropologists, social scientists, and user researchers to assess and improve this technology. The company also has an Ethics AI Committee that reviews AI systems based on ethical and responsible principles, including potential biases, limitations, and possible harmful use cases. This interdisciplinary approach helps ensure that AI technology, such as deepfake detection, benefits humanity without causing harm. Demir said, "We have legal experts, social scientists, psychologists, all coming together and pointing out limitations, looking for biases - algorithmic biases, systemic biases, data biases, any kind of biases." The team scans the code to identify "any potential use cases of technology that could harm people." As deepfake technology becomes more prevalent and sophisticated, the development and implementation of detection technology to combat misinformation and other harmful consequences become increasingly important. Intel Labs' real-time deepfake detection technology provides a scalable and effective solution to this growing problem. By incorporating ethical considerations and collaborating with experts in various fields, Intel is committed to creating a future where AI technology is used responsibly and benefits society. Source: https://www.zdnet.com/article/real-time-deepfake-detection-how-intel-labs-uses-ai-to-fight-misinformation/

RECENT AI NEWS

RECENT AI TOOLS