IMAGE: X-PHY Inc.
As deepfake technology advances at an alarming rate, the need for robust detection tools has never been more critical. X-PHY Inc., part of Singapore’s Flexxon Group, has risen to the challenge with its Deepfake Detector.
The X-PHY Deepfake Detector is a real-time deepfake detection tool that empowers users to verify the authenticity of videos, audio and images directly on their devices, without relying on the cloud. The company is currently demoing Deepfake Detector at RSA Conference (RSAC) 2025.
With deepfake content growing exponentially, the race to provide protection for corporates and consumers is heating up. On social media alone, deepfakes grew 550 per cent between 2019 and 2023, making the need for effective detection tools more urgent than ever. Even online romances are getting not spared – deepfake romance scams raked in US$46 million from men across Asia.
X-PHY’s innovation is designed to combat AI-generated deception, enabling users to verify the authenticity of digital media including videos, images and audio in real time.
Camellia Chan, CEO and co-founder of X-PHY Inc., explained: “The X-PHY Deepfake Detector is powered by AI models trained specifically to catch synthetic media in real time. It uses several types of deep learning architectures, advanced AI models to detect visual irregularities, and long short-term memory networks (LSTMs) and gated recurrent units (GRUs) to track speech and motion patterns. These models work together through a “zero trust” scoring engine that flags anything suspicious.”
X-PHY tested the tool on large datasets of generative adversarial network (GAN)-generated deepfakes, including content from platforms like StyleGAN and FaceSwap, achieving up to 90 per cent detection confidence in internal controlled lab tests.
In a demo (see below), X-PHY’s presenter used a visual deepfake site to masquerade as an Asian celebrity on a Zoom video conference, and the software flagged his video as fake on the screen.
Run anywhere
Upon activation, the X-PHY Deepfake Detector uses a multi-modal AI to analyse video, image and audio streams in real time. By examining facial micro-expressions, voice fingerprints, and (GAN)-generated artifacts, it flags signs of manipulation – even across multiple video windows. Detection is performed entirely on-device, preserving privacy and functioning even without an internet connection.
This is achieved through the Deepfake Detector’s use of advanced temporal and spatial AI analysis, powered by pre-trained neural networks. These models are capable of identifying subtle inconsistencies across facial movements, audio waveforms, and image artifacts – common signs of AI-generated content.
Chan points out that less than 5 per cent of enterprises today have deepfake detection protection, attributing this seeming indifference to user perception of the software being cumbersome, burdensome and unintuitive.
“We are confident that what we have here will provide a lightweight, frictionless and privacy-first solution that empowers users to detect deepfakes seamlessly, without needing to change the way they work,” she added.
Zero trust approach
Built on zero trust principles, the solution adds an additional layer of authentication and verification at the device level, helping organisations strengthen their cyber resilience against AI-powered deception and reducing reliance on external validation systems that often introduce unnecessary operational complexity.
The solution is application-agnostic, compatible with leading platforms like Teams, Zoom, Webex, Chrome, YouTube and Meta. Users can activate it with a single click when joining a meeting, where it runs autonomously for a preset duration and can be re-engaged as needed.
X-PHY has built strong relationships across the cybersecurity ecosystem, including with global tech partners, platform providers and public-sector agencies. Its collaborations span OEM partners for endpoint security, innovation centres like BWTech@UBMC in the U.S., and major government bodies.
Commercially, X-PHY works with PC manufacturers including Lenovo, ASUS and HP, which integrate X-PHY technology into their laptops. These partnerships bring real-time, hardware-based cybersecurity to users right out of the box.
In a world where digital deception is on the rise, X-PHY’s Deepfake Detector offers a vital solution. By providing a lightweight, privacy-first tool that seamlessly integrates into existing workflows, X-PHY empowers users to stay one step ahead of AI-generated threats.
