0
A screenshot released by the Singapore Police Force shows a deepfake Zoom video call impersonating Prime Minister Lawrence Wong, highlighting a recent wave of sophisticated AI-driven financial scams in the country. / Photo courtesy of the Singapore Police Force

Sophisticated scams leveraging artificial intelligence to clone the faces and voices of national leaders are sweeping through Southeast Asia. In Singapore, a high-profile deepfake impersonating the prime minister recently appeared in a Zoom video conference, swindling a victim out of 4.9 million Singapore dollars (approximately 5.7 billion Korean won), drawing widespread concern.

A Singaporean citizen recently joined a private video conference after receiving a WhatsApp message purportedly sent from the Cabinet Secretariat, the South China Morning Post (SCMP) reported on July 12. The screen displayed Singaporean Prime Minister Lawrence Wong alongside the president, cabinet ministers, and foreign diplomats discussing financial support related to the Strait of Hormuz.

After the deepfake prime minister concluded the meeting and thanked the victim for attending, a fake lawyer took over the call to provide wire transfer instructions. Believing it to be a legitimate government procedure, the victim transferred 4.9 million Singapore dollars. The receiving account vanished immediately afterward.

The weaponization of AI to impersonate political figures is not confined to Singapore. In Indonesia, shortly after Prabowo Subianto was elected president in 2024, an Instagram video cloning his face and voice went viral. Duped by the fake prompt asking if anyone had yet to receive their government subsidy, unspecified individuals across 20 provinces transferred up to 1 million Indonesian rupiahs (about 82,500 Korean won) each in administrative fees.

In the Philippines, a dentist lost 93 million pesos (around 2.25 billion Korean won) after falling for an online investment advertisement that superimposed the face of President Ferdinand Marcos Jr.

Interpol identified AI-driven crime as the top emerging threat in Southeast Asia in a report published in June. Between January 2024 and March 2025, phishing and related fraud accounted for the highest financial losses among cybercrimes in the Asia-South Pacific region, with one-third of surveyed countries recording over 10,000 cases. The same report noted that deepfake-related discussions on cybercrime forums and Telegram surged by 600% over a five-month span in 2024.

Experts point out that these scams succeed not due to technical vulnerabilities, but by exploiting human psychology. Interpol highlighted that the core of deepfake fraud relies on social engineering—targeting human trust rather than system flaws.

"Because many people are conditioned to respect authority, impersonating a trustworthy figure triggers compliance rather than suspicion," explained Jun Angelo Suglao, a clinical psychologist with the Psychological Association of the Philippines. The situation worsens when voice cloning technology is introduced. "When you hear the crying voice of a loved one pleading for help, rational judgment completely paralyzes," Suglao added.

AI is transforming not only how victims are targeted, but also how syndicates operate. Jeremy Douglas, a regional deputy director for the United Nations Office on Drugs and Crime (UNODC), noted that criminal organizations are now operating scams with fewer personnel, shifting from massive, localized scam compounds to decentralized, network-based hubs.

"Previously, people could spot scams through grammatical errors, but now AI generates flawless prose and counterfeit websites," said Art Samaniego, co-founder of Scam Watch Pilipinas. "AI isn't replacing criminals; it is doubling or tripling their efficiency."

Concerns are growing that traditional detection methods are failing to keep pace with these rapidly evolving AI tactics. "As AI produces human-grade output, classic detection cues like awkward phrasing or unnatural eye blinking have become obsolete," Samaniego pointed out.

According to a Microsoft report, phishing emails crafted by AI yield a 4.5 times higher click-through rate than traditional ones. Samaniego advised that when receiving urgent or large-scale financial requests, individuals must never respond directly to that specific message or call—even if it appears to come from an acquaintance—and must verify the sender's identity through a separate, independent channel.

"Seeing or hearing someone’s face and voice online is no longer proof of authenticity," Samaniego emphasized, calling for proactive public education before financial damage occurs.

                                                                                                              Jeong Ri-na