ASEAN updates guidance on deepfakes

The threat of deepfakes is entering high-level discussions from Southeast Asia to Davos.

The Association of Southeast Asian Nations (ASEAN) has launched an expanded guide on governance and ethics for generative AI, including voluntary policy recommendations on risks and responsible use.

The guide, which builds on a 2024 document, explains that the technology poses six major risks: Deepfakes, intellectual property (IP) infringements, embedded biases, factual inaccuracies as well as issues with privacy and confidentiality, mistakes and anthropomorphism. It also explains frontier and systemic risks linked to the long-term evolution of advanced generative AI models, according to the Phnom Penh Post.

The updated document was released at the ASEAN Digital Ministers’ Meeting 2025 held in Bangkok, Thailand this week. During the event, the organizations also published the Bangkok Digital Declaration which pledges to boost regional investment into emerging technologies such as AI, boost collaboration on online scams and the free flow of data within the region.

Cybersecurity firms warn against deepfake threat

On the sidelines of the World Economic Forum (WEF) in Davos, Switzerland, cybersecurity industry representatives also shared their views on the effects of deepfakes.

Cybersecurity attacks are on the rise, says Check Point Software President Rupal Shah Hollenbeck. But while 67 percent of organizations in the company’s survey believe that AI will have the largest impact on their security, only 30 percent are doing anything about it.

“A part of it is because they don’t know what to do,” she says. In the meantime, generative AI tools are becoming more sophisticated and the bad guys are becoming faster.

Check Point’s survey has also shown that cyberattacks have risen 44 percent from 2023 to 2024. An increasing number of them are ransomware, which saw a 90 percent year-on-year rise, Shah Hollenback adds.

Similar warnings on deepfakes are coming from Paravision. In a new blog post, the AI vision company says that facial recognition technology argues that traditional facial recognition, which cannot detect whether a face is real, is not enough to fight against fraud and unauthorized access.

Cyberattackers are constantly developing new ways to create fake faces and subvert identity systems with the help of generative AI, deepfakes and synthetic images. This means biometric authentication and identification developers must add multiple layers of capabilities such as liveness and deepfake detection.

In addition, these features should be complemented with other fraud-preventing technologies. Synthetic identities can include fabricated information such as addresses or credit histories, requiring solutions that will also perform document and data validation.

“Where one technology may falter, the other picks up the slack, ensuring that fraudsters can’t bypass security measures, no matter how sophisticated their techniques,” says Paravision.

Article Topics

ASEAN  |  biometric liveness detection  |  biometrics  |  cybersecurity  |  deepfake detection  |  deepfakes  |  fraud prevention  |  Paravision  |  World Economic Forum