Thailand Royal Police have deployed machine learning algorithms and national surveillance networks to detect and disrupt deepfake scams, a move aimed at reducing financial losses from synthetic media fraud currently exceeding ฿25 billion annually. As criminal networks exploit generative AI to impersonate officials, relatives, and financial executives, law enforcement is racing to outpace the technology they are chasing.
Why This Matters:
• Working-age residents (20–49 years old) remain the primary targets, accounting for the largest share of victims recorded in 2024, with losses averaging significantly higher per case as the scams evolve.
• The "Police Care" app now includes a real-time officer verification tool and direct connection to freeze suspected mule accounts within minutes—critical for anyone receiving unexpected financial demands.
• Criminal syndicates are cloning voices and faces of family members and law enforcement to bypass banking KYC systems and convince victims to transfer funds.
The AI Arms Race: Detection Versus Deception
The Thailand Cyber Crime Investigation Bureau (CCIB) has adopted AI-driven surveillance and machine learning detection tools designed to identify anomalies in digital video and audio files. These systems scan for unnatural facial movements, inconsistent voice patterns, and metadata irregularities that betray synthetic manipulation. The Intelligent Bird Eye Operation Centre (IBOC), originally deployed for real-time security monitoring in economic zones and tourist areas, now feeds data into broader cybercrime investigations, enabling authorities to flag suspicious deepfake content before it circulates widely.
Law enforcement also relies on the SHIELD database—a cross-border intelligence platform linking financial transaction data, suspect profiles, and syndicate networks across Southeast Asia. SHIELD integrates with the Anti-Cyber Crime Operations Centre (AOC 1441), a 24-hour hotline launched in 2023 that processes fraud reports, issues account freezes, and coordinates compensation claims. The platform's AI capabilities analyze patterns in reported scams, helping officers anticipate new attack vectors and deploy countermeasures proactively.
Yet detection remains a moving target. Throughout 2024, Thailand Royal Police warned that syndicates had begun using AI to impersonate officers themselves, complete with fabricated video calls and official-looking documentation. Recently, Thai PBS Verify flagged a surge in "AI Clone & Deepfake" schemes targeting families: criminals harvest social media photos and voice recordings, then generate convincing distress calls or ransom demands using the likeness of a relative.
High-Value Targets and Financial Toll
According to AOC 1441, online fraud in Thailand has generated hundreds of thousands of victims and billions of baht in losses in recent years. Young professionals aged 21–30 have emerged as increasingly vulnerable targets. This group's frequent use of digital banking, cryptocurrency platforms, and e-commerce makes them lucrative marks for syndicates deploying face-swap technology to defeat biometric authentication.
Women, particularly those over 50, also feature prominently in victim profiles. Romance scams using deepfake videos of fictitious suitors have extracted millions from elderly Thai women, exploiting trust-building conversations conducted over weeks or months. Meanwhile, crypto investors remain at risk from fake endorsement videos featuring prominent business figures or celebrities—content that appears authentic down to lighting, gestures, and speech cadence.
The Department of Special Investigation (DSI) has issued advisories highlighting a persistent tactic: syndicates use deepfake face animation to impersonate legitimate account holders during video-based KYC verification on banking and digital wallet platforms. By feeding synthetic video streams into live camera feeds, criminals open mule accounts that launder stolen funds before authorities can trace them.
Legal Framework: Patchwork Enforcement
Thailand has not enacted deepfake-specific legislation. Instead, prosecutors rely on the Computer Crime Act (2007), which criminalizes the insertion of false data into computer systems likely to cause public harm, and recent amendments strengthening penalties for technology-enabled fraud. These measures target operating mule accounts, mandate SMS filtering for fraudulent messages, and impose joint liability on banks, telecom providers, and online platforms that fail to meet cybersecurity standards.
Thailand's Anti-Fake News Center has clarified that using AI deepfakes to alter another person's image or voice in a manner causing embarrassment or reputational damage violates both the Computer Crime Act and the Criminal Code's defamation provisions. Convictions carry imprisonment and fines, though enforcement remains inconsistent, particularly when perpetrators operate from overseas call centers.
Police Care App: Real-Time Verification and Account Freezing
The "Police Care" mobile application represents one of Thailand's most practical anti-fraud tools currently available. Its "Verify Officer" feature allows residents to confirm whether someone claiming to be police is legitimate—users input the officer's name or scan a QR code badge to cross-reference against the national database. This counters impersonation scams in which criminals pose as investigators demanding immediate payment to resolve fabricated legal issues.
The app also connects directly to the Thai Bankers' Association network. When a user reports a suspected scam, Police Care can issue a temporary freeze on the recipient account within minutes, preventing further withdrawals while investigators assess the case. This rapid-response mechanism has recovered millions of baht that would otherwise vanish into layered money-laundering chains.
Regional Comparison: Thailand's Position in Southeast Asia
Singapore leads the region with targeted deepfake legislation. Its Elections (Integrity of Online Advertising) (Amendment) Act imposed fines up to SGD 1 million for platforms hosting election-related deepfakes, while the Online Criminal Harms Act (OCHA) grants authorities power to issue takedown orders and restrict accounts linked to fraud. Additional online safety measures are being implemented to protect victims from AI-generated abuse.
Vietnam has enacted comprehensive AI governance via Decree 142, which mandates risk classification for generative AI systems, requires deepfake labeling, and bans synthetic media used for deception. Platforms must monitor and remove prohibited content or face penalties.
The Philippines has proposed the Deepfake Regulation Act (House Bill 10567), which would impose prison terms and fines on creators and distributors, plus penalties for platforms that refuse to delete flagged content. The country also blocked access to certain AI chatbots following their misuse in generating explicit deepfake images.
Malaysia is developing comprehensive AI legislation and has passed the Online Safety Act, which targets financial fraud and cyber harassment. The government is working to assign social media platforms responsibility for content moderation.
Indonesia relies on its Electronic Information and Transactions Law (UU ITE) and revised Criminal Code (KUHP) to prosecute deepfake-related offenses. The Ministry of Communication and Information (Kominfo) has issued ethical guidelines for AI use and participates in regional coordination against harmful deepfake content.
At the regional level, ASEAN released an updated Guide on Governance and Ethics of Generative AI in 2025, offering voluntary policy recommendations on deepfake risks. Singapore has also developed an ASEAN Scam Prevention Toolkit, focusing on call and SMS screening, SIM card controls, and data sharing.
What This Means for Residents and Expats
If you receive an unsolicited video call or voice message from someone claiming to be family, a colleague, or law enforcement, assume it may be synthetic until verified through a separate communication channel. Do not rely solely on visual or audio cues; deepfakes now replicate micro-expressions, accent patterns, and background noise convincingly.
Download the Police Care app and use the officer verification feature before complying with any financial demand from someone identifying as police. If you suspect fraud, report immediately via AOC 1441 to trigger account freezes and investigative protocols.
For those managing digital wallets or cryptocurrency accounts, enable multi-factor authentication beyond biometrics. Face and voice recognition alone can be defeated by advanced deepfakes; pairing these with hardware tokens or SMS codes adds a critical layer of defense.
Employers and financial institutions should implement liveness detection during KYC processes—systems that require users to perform random gestures or respond to prompts that pre-recorded deepfakes cannot replicate. As syndicates refine their tools, static video verification becomes increasingly unreliable.
The Technology Paradox
Thailand Royal Police have also used deepfake technology offensively, creating synthetic public service announcements to warn citizens about emerging scams. This dual-use reality underscores the paradox at the heart of the AI arms race: the same generative models that produce fraud also power the detection systems meant to stop it.
Machine learning algorithms trained on thousands of deepfake samples can spot telltale artifacts—flickering edges, mismatched lighting, or audio desynchronization—that human reviewers miss. But as models improve, these artifacts disappear. Some law enforcement analysts have acknowledged that maintaining detection rates presents an ongoing challenge as criminal AI tools approach production-studio quality.
International cooperation remains essential. ThaiCERT has reported that Thailand participated in coordinated takedowns of deepfake pornography websites involving international cyber police and law enforcement partners. Such operations demonstrate that while Thai authorities can disrupt domestic syndicates, the underlying technology and hosting infrastructure often reside beyond national jurisdiction.
The Road Ahead: Legislation and Public Awareness
Legal experts and civil society groups have called for Thailand to follow Singapore and Vietnam in enacting deepfake-specific statutes that clearly define prohibited uses, establish platform liability, and mandate disclosure labels on synthetic media. The existing patchwork of computer crime, defamation, and data protection laws leaves gaps that savvy defendants exploit, arguing that their actions fall outside narrow statutory language.
Meanwhile, the Ministry of Digital Economy and Society (MDES) continues public awareness campaigns emphasizing common scam tactics: impersonation calls using AI voices, fake investment platforms endorsed by deepfake celebrities, romance scams with synthetic video, and KYC fraud via face-swapped verification. The ministry's research indicates that informed residents are significantly less likely to fall victim, underscoring the value of education alongside enforcement.
As generative AI becomes more accessible—open-source models now enable deepfake creation on consumer laptops—Thailand's cyber police face an opponent that scales faster than budgets or personnel. The challenge facing Thai authorities is whether detection technology and legal frameworks can keep pace with evolving synthetic fraud threats, or whether the next generation of attacks will outrun existing countermeasures.
