McAfee at CES 2024: Deepfake Audio Detection Technology using AI

According to a report, 84% of Americans are concerned about how deepfakes will be used in 2024¹. Deepfake audio or deepfake voice scams have become a pernicious challenge for people who post their images, videos, and voices online, mostly on social media platforms. Almost every digital image or video published online is a deepfake, or on the verge of becoming one. Out of 1.8 billion images and videos uploaded online, nearly 50% are digitally manipulated for “benign reasons” using morphing or enhancement filters for magazine covers. It becomes a problem when these crossover to deepfakes used for propagating misinformation or worse, used as a weapon against a government or nation by threat actors².

Deepfake scams are engineering using generative AI or GenAI tools for cloning voices, images, and videos. Also called “cheap fakes”, the problem of manipulating personal identity using AI and deep learning can create larger problems for the whole world in 2024. Threat actors use Generative adversarial networks (GANs) to create deepfake videos and images³.

Elections are planned in major democracies of the world, and the use of GenAI for creating deepfake audio assets could sway the results completely. Government institutions are aware of this problem, and so are highly regulated industries such as banking and insurance, education, healthcare, education, and military services. But, what are the steps taken to stop cybercriminals from accessing these deepfake creation tools?

McAfee Reveals 2024 Cybersecurity Predictions: Advancement of AI Shapes the Future of Online Scams

With open source and free access to genAI tools, stopping cyber criminals and scammers from using AI for creating deepfakes is nearly impossible. However, it’s somewhat possible to detect AI-generated deepfake content. And, at CES 2024, the world’s most advanced deepfake audio detection technology was on display. McAfee, a leader in the device and identity protection software industry, announced its ambitious Project Mockingbird. McAfee admitted that its new AI-driven deepfake detection tool has an accuracy of more than 90% in detecting and exposing maliciously created audio in videos. As we head into a challenging year in 2024, users can rely on McAfee’s Project Mockingbird to navigate their ever-evolving digital world. At CES 2024, we spoke to McAfee’s SVP and Chief Technology Officer, Steve Grobman about the role of AI ML tools in disrupting the security landscape. The latest Project Mockingbird deepfake audio AI technology puts the onus on the engineering world to develop a comprehensive portfolio of AI models that can safeguard the interests of an individual, community, and nation.

Here’s the excerpt from our full interview with Steve.

Hi Steve, welcome to our AiThority.com Interview Series. Please tell us a little bit about your role and responsibilities at McAfee. How did you arrive at the company?

Thanks so much for the opportunity! As McAfee’s Chief Technology officer, I’m responsible for setting the technical strategy and direction for the company and leading our research and development efforts. My team constantly monitors the evolving threat landscape and leverages new and emerging technologies to develop innovative solutions to protect consumers from today’s threats.

McAfee Report Details How Advancement in AI is Creating New Mobile Threats, Aiding Scammers

Before I joined McAfee, I spent more than two decades at Intel in several leadership positions related to cybersecurity.

With the massive proliferation of AI ML in the business landscape, could you tell us more about the changing expectations from top organizations?

The new wave of Advanced AI technology simultaneously lowers the barrier to entry and enhances the creative capabilities of individuals and organizations. From a cybersecurity perspective, the latest generative AI tools make developing convincing scam and phishing emails easier. For example, cybercriminals can use generative AI to create convincing messages free of spelling and grammar mistakes, eliminating one of the red flags traditionally used to identify fraudulent messaging. Additionally, bad actors can create fake images, videos, and audio content, all of which can act as a powerful tool of influence in the latest generation of scams.

The good news is that McAfee is leveraging advanced AI to detect the latest threats and provide personalized recommendations on how consumers can stay safe in our evolving digital world.

You are at CES 2024. What would be your key mission at the event?

My key mission at CES 2024 is to continue educating partners and consumers about the great work that McAfee is doing and reinforce our mission to protect consumers and their digital lives. This year, we’re unveiling Deepfake Audio Detection technology created by our research and development team. In short, we’ve developed advanced AI-detection capabilities that will help consumers discern what’s real and what’s fake, proactively alerting and protecting them against malicious, AI-generated content. 

Deepfake audio is enabling criminals to poach victims in different scams and frauds. Could you tell us how these deepfake audio scams impact lives?

Cybercriminals are increasingly using fabricated, AI-generated audio to create highly convincing scams that rob people of money and personal information, enable cyberbullying, and manipulate the public image of prominent figures. Our data tells us that 70% of Americans are concerned about deepfakes making it hard to trust what they see and hear online.

Please tell us about your deepfake detection technology and how it works.

McAfee’s AI-powered Deepfake Audio Detection technology, known as Project Mockingbird, uses AI-powered models to identify whether audio is AI-generated or authentic. Often, fake audio is part of a video that is either a “deep-fake” — where a person appears to say something they have not — or a “cheap-fake” where audio is overlaid on generic video or the narration of unrelated video content. With an accuracy rate of more than 90%, our technology can detect AI-generated content across a wide range of scam scenarios including deepfake-driven cyberbullying, reputation manipulation, and investment scams.

How are you preparing for the gen AI evolution and Edge evolution at McAfee yourself? Are there any key milestones from 2023 that you would like to share with the IT leaders?

At McAfee, we’ve always had an eye to the future, leveraging our foresight and experience to anticipate what’s next. More than five years ago, in a keynote at RSA, I predicted the rise of deepfakes and AI-generated malicious content. But, we didn’t just predict this reality; we prepared for it. This is evident in our advancements and speed to deliver compelling technologies to address today’s evolving threat landscape. We are committed to winning the ‘cat-and-mouse’ game that we play with the bad actors and have confidence that our AI can beat their AI.

Thank you, Steve! That was fun and we hope to see you back on AiThority.com soon.

[To share your insights with us, please write to sghosh@martechseries.com]

References

1. McAfee’s December 2023 Deepfakes Survey
2. Akhtar Z. Deepfakes Generation and Detection: A Short Survey. Journal of Imaging. 2023; 9(1):18. https://doi.org/10.3390/jimaging9010018
3. Almars, A. (2021) Deepfakes Detection Techniques Using Deep Learning: A Survey. Journal of Computer and Communications, 9, 20-35. doi: 10.4236/jcc.2021.95003.