AI-Driven Scams Cause Widespread Financial Harm in Taiwan
Scam operations in Taiwan are increasingly using AI to generate synthetic content and automate social engineering attacks, with reported annual losses exceeding 9 billion Taiwan dollars. Scammers route victims from social media platforms to encrypted messaging apps and use AI-generated deepfakes and realistic dialogues to build trust before extracting money through fake shopping, investment, and phishing schemes.
What happened
Trend Micro's 2026 fraud trend prediction report indicates AI is being used by scam groups at scale to generate realistic synthetic people, dialogue and content, and can conduct research and adjust strategies without human intervention.
Over 70% of SMS scam victims were requested to switch communication platforms, and most scam websites disappeared within 24 hours. Common scams originate on Facebook and Instagram, then redirect victims to LINE and other encrypted messaging apps where security systems have difficulty detecting malicious links.
Common scams include online shopping fraud promoted as limited-time offers with below-market prices, phishing SMS impersonating logistics companies, immediate payment app fraud impersonating banks or employers, and fake brand and influencer scams. Nearly 50% of victims discovered they were scammed only after being notified by banks or police.
Reported impact
A 165 anti-fraud dashboard shows approximately 600 fraud reports daily in Taiwan in the year preceding January 2026, with case amounts exceeding 200 million Taiwan dollars. Scam losses reportedly exceed 9 billion Taiwan dollars annually.
Over 90% of Taiwanese residents reported concern about friends and family being scammed. Over 60% of residents had seen scam messages impersonating known brands, yet over 40% would still shop on unfamiliar websites, and only 13% of victims reported incidents to police.
Index analysis
Scam methods are evolving from one-time attacks to long-term trust cultivation schemes called '養套殺' (raise-set-kill), with multi-channel attacks across social media, SMS, voice and video. Over 40% of residents have begun using AI to assist with shopping or booking, potentially increasing fraud risk in high-reliance online decision scenarios.
Governance controls
Sinopac Bank deployed AI facial recognition at ATM entrances to prevent masked fraudsters, introduced a dedicated SMS shortcode '68807' for message authentication, and launched a URL verification tool to identify phishing websites. The bank's AI models detect suspicious account behavior and block automated transactions, achieving a 51.16% fraud prevention rate by end of 2025.
Sinopac Bank implemented cross-unit fraud prevention teams, transaction limit controls, biometric authentication, and device binding mechanisms. The bank conducts public awareness campaigns and collaborated with police through memoranda of understanding and information sharing to identify suspicious transactions across institutions.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.