Pioneering deepfake detection and AI research at AIRO

These hyper-realistic fake videos and images, powered by artificial intelligence, pose significant ethical and security challenges.

Torrens University’s Artificial Intelligence Research and Optimisation Centre (AIRO) is at the forefront of addressing these concerns, making strong advancements in the detection and mitigation of deepfake content.

Professor Tony Jan, Centre Director of AIRO at Torrens University, is fronting this valuable research and says the ability to detect these fakes is an ongoing challenge, especially as the technology creating them becomes more sophisticated.

“Deepfakes are a significant concern,” Professor Jan explains. “Fake videos can deceive the public, spread misinformation, and cause irreversible harm to individuals and organisations.”

The threat of deepfakes

The dangers of deepfake technology are multifaceted, ranging from personal to global. High-profile personalities and organisations are prime targets, with deepfakes often used to impersonate them in politically sensitive scenarios. More sinister applications include the creation of fake explicit content, which can lead to defamation, extortion, and severe emotional damage to victims.

“It’s not just celebrities or public figures who are at risk,” Professor Jan warns. “With enough publicly available images, anyone can become a victim. This is particularly concerning in cases where teenagers’ or children’s images are misused, leading to unimaginable consequences.”

Advancing deepfake detection

The AIRO team at Torrens University in collaboration with peers, have outlined a novel approach to deepfake detection combining advanced machine learning techniques and data fusion methods. The study, involving datasets of fake videos, highlighted subtle but consistent differences between real and synthetic facial expressions that can only be identified by artificial intelligence.

“Our proposed solution used the efficient artificial neural network to achieve one of the highest accuracies in detecting deepfakes. While humans struggle to identify these subtle inconsistencies, advanced AI systems can detect them with remarkable precision,” said Professor Jan.

Professor Jan says such a solution can have a positive impact on the community, as well as future commercialisation opportunities.

Ethical considerations and potential benefits of deepfakes

Despite the malicious intent often associated with deepfakes, the underlying technology has potential for positive applications.

Professor Jan envisions heartwarming uses, such as creating messages from loved ones for significant life events.

“Imagine someone nearing the end of their life recording messages for their children, to be shared on birthdays or special occasions,” he suggests. “The same technology that poses a threat could be used to bring comfort and connection.”

Protect yourself against deepfakes

Professor Jan’s advice to the public is clear: be cautious with what you share online.

“Avoid posting high-resolution images or videos of your family, particularly children, on social media,” he advises. “Even with privacy settings, social media platforms are vulnerable, and once content is online it’s difficult to control where it goes.”

As well as providing cybersecurity education that will help shape industry leaders, AIRO is encouraging greater awareness and legal measures to combat the misuse of deepfake technology.

“While there have been defamation lawsuits, the legal landscape is still catching up to these advancements,” Professor Jan notes. “We must prioritise robust regulations and public education to protect individuals and society at large.”

Leading the way in artificial intelligence (AI) research

As AIRO continues to refine its detection technology, Torrens University is cementing its position as a global leader in AI research. The team's ground-breaking work not only addresses immediate threats but also lays the foundation for ethical applications of AI technology. Their efforts exemplify the university’s commitment to innovation and societal impact.

This research knowledge also supports teaching and learning in the new cybersecurity courses, so students learn the latest skills and knowledge in digital forensic and data security placing them in a great position for future employment.

In an age where seeing is no longer believing, AIRO’s research provides hope for a safer, more trustworthy digital future. The challenges are immense, but with researchers at Torrens University leading the charge the battle against deepfakes is well underway.