In the 1970s, psychologist Paul Ekman embarked on a groundbreaking journey to explore how facial expressions might reveal underlying emotions and intentions. His research focused on microexpressions—those fleeting, involuntary facial movements that can signal a person’s true feelings. This work has since become vital in various fields, particularly as society grapples with the emergence of deepfake technology.
Deepfakes, which are artificially generated videos that convincingly mimic real people, pose a significant threat to personal security and integrity. They are increasingly exploited for identity theft, fraud, and extortion. To counter this alarming trend, European law enforcement agencies are now harnessing Ekman’s insights into microexpressions, with the help of Herta Security, a company specializing in facial biometrics.
Herta Security’s Innovative Detection Method
Based in Barcelona, Herta Security is taking a fresh approach in the fight against deepfakes. Rather than relying on traditional detection methods, which can quickly become outdated as technology evolves, Herta focuses on identifying microexpressions that deepfakes often fail to replicate.
“We examine deepfake detection from a human perspective, utilizing microexpressions that current deepfake technology struggles to mimic,” says Manuel Pastor, Herta’s Chief Technology Officer. He notes that these artificial videos often lack natural blinking patterns and authentic smiles, making them easier to identify.
Herta’s strategy sets it apart from existing tools, aiming to develop technology that can adapt to future advancements in deepfake creation. By focusing on the subtle involuntary movements of the face, the company seeks to create a robust solution that stands the test of time.
EITHOS: A Collaborative Effort Against Identity Theft
Herta Security is a key member of the European Identity Theft Observatory System (EITHOS), a consortium comprising 12 organizations dedicated to combatting identity theft across Europe. Funded by the EU’s Horizon 2020 initiative, EITHOS unites universities, law enforcement, and private firms in a shared mission to detect and combat fraudulent identities.
The timing of this initiative is critical. As deepfake-related financial fraud is expected to surge, according to a Deloitte poll, and instances of non-consensual deepfake pornography continue to rise, the need for effective detection tools is more pressing than ever. EITHOS is developing a suite of four tools to aid law enforcement, including deepfake video detection led by Herta and deepfake audio detection coordinated by Vicomtech, a Spanish AI research center.
The Mechanics of Deepfake Detection
Herta’s detection software compares facial expressions from suspected deepfake videos with verified recordings of real individuals. By analyzing involuntary expressions, such as how often a person blinks or smiles, the software seeks to uncover inconsistencies that reveal a video’s authenticity.
“When someone speaks naturally, their blinking and smiling occur at specific frequencies that are hard for deepfakes to imitate,” Pastor explains. Herta’s technology employs advanced machine-learning algorithms, drawing on the company’s experience with microexpression analysis tools previously used by the Spanish National Police.
Current Developments and Future Goals
Currently in the beta testing phase, Herta’s deepfake detection software is being evaluated by law enforcement agencies in Sweden, Greece, and Spain. The company anticipates that its product will soon be available to police forces throughout Europe.
“Our ultimate goal is to create a commercial product primarily for law enforcement agencies in Europe and beyond,” Pastor states. EITHOS, which started in 2022, aims to deliver its comprehensive tools by 2025, marking a significant milestone in the fight against identity theft and fraud.
Raising Awareness on Identity Theft
In addition to technological advancements, EITHOS places a strong emphasis on educating the public about identity theft and data protection. Partnering with institutions like the University of Bologna and the Technical University of Madrid, the consortium studies the various social and psychological dimensions of identity theft.
Pastor emphasizes that many victims often feel too embarrassed to report their experiences to authorities. “They usually keep it to themselves, trying to resolve the issue privately,” he says. Identity theft can affect anyone, including those who are tech-savvy and educated, often manifesting as extortion through fake pornographic videos that leave victims feeling trapped.
“It’s crucial for citizens to understand that they have access to technical tools and support from law enforcement for these crimes,” Pastor asserts.
