How HTX Is Winning the War Against AI Deepfakes

At the recent INTERPOL Conference on AI in Digital Forensics in Tokyo, Singapore's Home Team Science and Technology Agency (HTX) demonstrated its significant progress in using artificial intelligence to combat sophisticated digital crimes. The event, held from October 28 to 30, 2025, brought together global experts from law enforcement, academia, and industry to address the challenges and solutions related to AI in digital investigations.

The Double-Edged Sword of AI

While advancements in AI are enhancing operational efficiencies for law enforcement, they also arm criminals with powerful new tools. The rise of deepfakes—hyper-realistic yet digitally manipulated videos and audio—synthetic identity fraud, and the use of generative AI (GenAI) to create malware present growing threats to public safety. These complex challenges require coordinated and technologically advanced responses from the global community.

To counter these threats, Singapore has developed a multi-pronged strategy that combines powerful AI-driven forensic tools with clear regulatory frameworks and robust legislation. This comprehensive approach was detailed by Ashley Roshan Colombo, a forensic examiner at HTX’s Digital & Information Forensics (DIF) Centre of Expertise.

Singapore's Defense Against AI-Driven Crime

Colombo highlighted the increasing frequency of GenAI-facilitated crimes in Singapore, referencing scams and a notable case involving the creation of deepfake nudes of students. He emphasized the critical need for AI tools that provide investigators with fast, user-friendly, and interpretable results, while also ensuring that forensic analysis methods remain explainable and reproducible for sound, incisive evidence.

Introducing AlchemiX A Powerful Deepfake Detector

Building on this, HTX speakers Ng Wai Hung and Priscilla Koh introduced AlchemiX, a deepfake detection system developed by HTX. This tool empowers investigators to adapt quickly to emerging threats and build internal expertise in analyzing synthetic media.

They demonstrated how AlchemiX goes beyond a simple yes-or-no verdict. The system detects even the most convincing deepfakes by identifying subtle physical inconsistencies, analyzing multiple faces within a scene, and assessing audio for temporal irregularities. This allows the tool to precisely highlight manipulated segments for investigators.

A Collaborative Future to Stay Ahead of Threats

The presentation showcased HTX's agility in iterating technology to address ever-evolving cybersecurity threats. Christopher Sia, Director of HTX's Sense-making & Surveillance (S&S) CoE, expressed his gratitude for the opportunity to share his team's work on a global stage. “We look forward to collaborating with agencies worldwide to stay ahead of deepfakes and emerging AI-enabled scams,” he said.

Lim Tuan Liang, Director of DIF CoE, echoed this sentiment, noting the growing synergy between AI deepfake detection and digital forensics. “HTX previously collaborated with INTERPOL under Project SynthWave, and this conference provides another valuable platform to deepen those partnerships,” he stated. “Looking ahead, the real opportunity lies in harnessing AI not just to detect deception, but to stay one step ahead, ensuring our tools evolve as rapidly as the threats they are designed to counter.”