The issue of deepfakes is gaining significant attention in Southeast Asia, where we are seeing both advancements and scams accelerate in real time. As deepfakes become increasingly accessible and realistic, they pose critical challenges to both individual users and larger organizations. These falsified audio and video recordings, crafted using sophisticated AI models, are now powerful tools that can be used to mislead, manipulate, or deceive. In a region where digital media consumption is high, the potential for misuse is alarming, especially as deepfakes begin to blur the line between reality and fabrication.

Tan Ah Tuan, Head of Ensign Labs at Ensign InfoSecurity emphasizes that deepfakes are not only a technical issue but also a societal one. According to him, the risks range from identity theft and fraud to the erosion of trust in public figures and institutions. Southeast Asia’s unique position as a rapidly digitizing region, coupled with its diverse cultural and political landscape, makes it particularly vulnerable to these emerging threats. Governments and enterprises are beginning to realize the seriousness of the situation, leading to increased focus on both legislative measures and technological solutions.

Here are 5 emerging cybersecurity threats to watch out for in Southeast Asia this year

One recent development in this space is Ensign’s launch of Aletheia, a real-time deepfake detection solution created by Ensign InfoSecurity. The tool can detect manipulated video and audio content with impressive speed and accuracy, offering a much-needed defence for enterprises, governments, and individual users. This solution arrives at a time when many are seeking effective methods to safeguard against malicious actors who use deepfakes to spread disinformation, engage in fraud, or discredit political figures.

We’ll dive deeper into how deepfakes are impacting Southeast Asia, featuring insights from Spokesperson A, who shares their perspective on how both businesses and individuals can navigate the growing threats posed by this deceptive technology. Through a series of questions, we’ll explore the challenges, solutions, and future outlook for combating deepfakes in this evolving digital landscape.

What are the most significant risks posed by deepfakes to enterprises and governments, and how does Aletheia address these threats in real time? Also, given that it performs analysis locally on users’ devices, how does this localised inference enhance privacy and performance compared to cloud-based solutions?

The most significant risks posed by deepfakes to enterprises and governments include financial fraud, misinformation, identity theft, and reputational damage. Deepfakes can be used to impersonate executives or government officials, leading to severe consequences such as unauthorised financial transactions, the leaking of sensitive information, and the manipulation of public trust. In some cases, deepfakes can fuel disinformation campaigns, resulting in national security issues or corporate crises.

Aletheia offers unparalleled protection against AI-driven manipulations in real-time

  • Unlike most existing solutions, which are forensic tools that analyse deepfake content offline (requiring users to record and upload suspected content to a web-based platform for analysis), Aletheia spots malicious deepfakes early – in real time – at the point of attempt, before users are compromised.   
    This is made possible by Aletheia’s deployment as an endpoint solution.
  • By analysing and detecting deepfakes locally on endpoints for real-time protection, Aletheia minimises latency issues and enhances privacy, ensuring that sensitive data is not sent to the cloud, which reduces the risk of breaches. 
  • Democratisation of AI to empower individuals: As an endpoint software or Chrome plug-in, Aletheia empowers individuals and employees to adopt a proactive defence against deepfakes, without the need for authorities’ or investigators’ intervention. It is easy to use and requires no additional training or change management efforts from companies.
  • Elevated digital security at the enterprise level: Aletheia can be integrated with the organisation’s security monitoring systems (e.g., Security Information and Event Management Systems).
  • When employees are exposed to deepfakes, Aletheia alerts cybersecurity analysts in real-time. By correlating these detections with other detections such as unusual behaviours or visits to phishing sites, automatic corrective, or preventive actions such as blocking access to these sites or temporarily reducing privileges of the affected user can be triggered immediately.  If multiple deepfakes are detected within a short period, a companywide mitigation can be swiftly executed. 

Could you explain the technical challenges in detecting hyper-realistic deepfakes, and how Aletheia’s AI-driven approach overcomes these obstacles? Also, can you walk us through Aletheia’s development process, particularly the role of open-source tools and training datasets in creating an effective deepfake detection system?

Detecting hyper-realistic deepfakes presents significant technical challenges. Advancements in AI, including increased access to open-source tools and more affordable GPU computing power, have made it easier for malicious actors to generate highly convincing deepfakes. As a result, threat actors can now produce sophisticated, hyper-realistic manipulations at scale, making detection increasingly difficult.

Aletheia overcomes these challenges through its AI-driven approach. The solution continuously evolves by training on open-source tools used to generate deepfakes and by leveraging vast datasets of authentic audio and video content. This diverse training helps Aletheia to identify a wide range of deepfake variations across both audio and video modalities, forming a suite of models. 

Aletheia is designed for real-time detection, with up to 90% accuracy, enabling it to spot anomalies across audio and video modalities. It is available as an endpoint software or Chrome extension. 

What steps are being taken to improve detection rates as deepfake technology evolves?

We continuously invest in R&D to stay ahead of the evolving cyber landscape and anticipate future cyber threats, including emerging deepfake trends. Through this proactive approach, we ensure that our detection solutions, such as Aletheia, are regularly updated to counter the ever-advancing capabilities of threat actors using deepfake technologies.

As deepfake detection currently relies on visual cues such as awkward facial movements and mismatched audio, how does Aletheia’s AI detect deepfakes that may not exhibit these traditional signs?

While traditional deepfake detection often relies on visible cues, Aletheia’s AI goes beyond surface-level signs. Our advanced algorithms analyse deeper, more subtle inconsistencies in both audio and video content. For video, Aletheia can detect minute anomalies in pixel patterns, lighting inconsistencies, or the unnatural blending of facial textures, elements that are not always visible to the naked eye. This allows it to identify deepfakes even when the manipulations do not exhibit obvious flaws.

For audio, Aletheia is trained to recognise subtle variations in speech patterns, intonations, and background noise that might go unnoticed in normal interactions but indicate potential tampering. Furthermore, Aletheia’s ability to analyse multiple layers of content simultaneously makes it particularly effective in more complex scenarios, such as live streams of political speeches, business interactions involving financial transactions or recorded footage. 

What’s next for Ensign InfoSecurity?

Over the next few years, Ensign InfoSecurity will be focusing on three areas of business growth.

First, is our continued investment in R&D, which is core to our business, Ensign will i) continuously improve our proprietary innovation suite to enable clients to detect and respond to cyberattacks with precision and speed, and ii) launch new AI-driven solutions to defend against emerging cyberthreats. 

The recent launch of Aletheia is one such example. We recognise that the issues of deepfakes transcend geographical, racial and ethnic boundaries. In this digital arms race where deepfakes generation and detection build on one another, we expect deepfake technologies to advance continuously. Aletheia will be continuously updated to stay ahead of the latest deepfake technologies and protect individuals from potential threats.  

We also plan to refine Aletheia based on the regions in which we operate. For example, Aletheia’s deepfake audio detection algorithm is being enhanced to support Asian languages and accents. 
The other area of focus is the expansion of our international footprint. Ensign operates out of 6 offices serving over 800 customers in 22 countries. To grow our market share organically, we endeavour to serve MNCs and LLEs beyond their headquarters and into other regions where they operate.  While our growth is primarily organic, judicious acquisitions will play an important role in further expanding our international footprint.