Humanity is under threat. Misinformation in the form of deepfakes is wreaking havoc in media, business, and geopolitics - eroding our basic sense of truth. The number of deepfakes is doubling every six months, and up to 90% of online content will be synthetically generated by 2026. 

To combat this threat, Intel had to transform its reputation from chip-maker to tech innovator, and reclaim its leadership position through revolutionary technology that enables global progress.

Through the first-ever real-time deepfake detection system, Intel reclaimed relevance by arming the world with an AI solution that certifies the truth and, moreover, certifies humanity.

The deepfake crisis has been exacerbated by generative AI—the sheer realism and volume of these videos are staggering. News outlets, social platforms, governments, and companies everywhere have no idea how to combat this new and complex issue.

Our challenge was to boost Intel’s reputation by highlighting how our Xeon processors and AI expertise help address the issue directly. Deepfakes rank as the most significant AI threat by tech researchers and policymakers. With these rapid advancements in generative AI, deepfakes are now imperceptible to the human eye

Intel's real-time deepfake detection demonstrates to the tech decision-maker target our expertise in high-performance processors and innovative solutions that bolster safety, security, and transparency in an age of misinformation.

Objectives:

• Reinvigorate Intel’s reputation as an innovator

• Drive engagement with tech decision-makers

• Open the door to new customers by highlighting the power of Xeon

• Establish a new standard for truth in media for businesses

Certified Human features a first-of-its-kind technology platform that detects deepfakes in real time by authenticating what makes us human. 

With the power of Xeon processors, Intel’s real-time detection can see what the eye cannot. While other deep learning-based detectors look at raw data to try to find signs of inauthenticity and identify what’s wrong with a video, Intel’s technology looks for authentic clues by accessing the subtle blood flow in the pixels of a video. When our hearts pump blood, oxygen levels in our skin tones change color ever so slightly—which even the most advanced deepfakes cannot replicate. These blood flow signals are collected from all over the face and the FakeCatcher algorithm translates these signals into spatiotemporal maps. Using Intel’s processors’ massive computing power, videos are analyzed at the pixel level, enabling AI and machine learning to instantly certify what’s human and what’s not.

Certified Human took years of experience across multiple teams in seemingly disparate disciplines from deep learning, video encoding, volumetric video production, to high-performance computing to develop.

From January 6th through September 22nd, 2021 a series of patents were filed for deepfake detection modeling by Ilke Demir, Senior Research Scientist at Intel. These patents would serve as the foundation for Intel’s real-time offering.

In November of 2022, we launched a content series telling the story of this new technology. To raise Intel’s profile and cultivate credibility among tech decision-makers, we leveraged the real voices of Intel's experts, like Demir, to feature the innovation across an ecosystem of organic content. From a series of short docs to gamified TikToks, live AMAs on Twitter Spaces, and Reddit forums, we engaged our B2B audience wherever they live online.

We then identified cybersecurity firms of some of the world’s most high-profile companies through customer engagement and product demos. Intel’s technology is now being deployed for clients like Visa, Microsoft, NATO, and the U.S. Department of Homeland Security.

With a new customer pipeline open, we’re preparing an integration with a major news media platform in advance of the 2024 Elections.

With the power of Xeon processors, Intel’s real-time detection can see what the human eye cannot. While other deep learning-based detectors look at raw data to try to find signs of inauthenticity and identify what’s wrong with a video, Intel’s technology looks for authentic clues by accessing the subtle blood flow in the pixels of a video. When our hearts pump blood, oxygen levels in our skin tones change color ever so slightly—which even the most advanced deepfakes cannot replicate. These blood flow signals are collected from all over the face and algorithms translate these signals into spatiotemporal maps. Using Intel’s processors' massive computing power, videos are analyzed at the pixel level, enabling AI and machine learning to instantly certify what’s human and what’s not.

Now that the tech development is complete, the mission turns to unlocking the most powerful applications. From social media platforms leveraging it to prevent users from uploading harmful deepfake videos, to global news organization using it to avoid inadvertently amplifying manipulated videos, and even nonprofit organizations employing the platform to democratize the detection of deepfakes for everyone. Our ambition is to make Certified Human available to every media platform, company, government, and human.

Intel's deepfake detection has proven to be a formidable weapon—with the power to analyze over 72 videos at the same time with 96% accuracy. Our content highlighting the technology produced 72% more organic shares per post and 56% more organic engagements among our social community.

Moreover, 88% of tech decision-makers now credit Intel for providing innovative solutions, directly deterring the perception that Intel only makes microchips. We also ignited the sales pipeline and opened new verticals via partnerships with cyber security firms. Conversations are actively happening with traditional and social media outlets to integrate this technology into everyday editorial operations to fight misinformation more responsibly and effectively.

Intel’s real-time deepfake detection provides an important step to staying ahead of generative AI and harmful deepfakes. A stronger foundation for rebuilding the very concepts of truth and trust. And a path forward to ensure we are all Certified Human.