In June 2025, Taiwan's Ministry of Digital Affairs (MODA) announced that six major platforms — Meta (Facebook and Instagram), Google (including YouTube), LINE, and TikTok — must publish annual fraud-prevention transparency reports by December 31, 2025. The mandate flows from the Fraud Crime Hazard Prevention Act (FCHPA), which took effect in July 2024. The requirements are specific and operational: platforms must verify advertiser identities using digital signatures or equivalent technologies, establish formal fraud prevention plans, remove or restrict fraudulent advertisements within 24 hours of notification, and disclose metrics covering fraudulent ad volumes, removal response times, and monthly active Taiwanese user counts. All four companies submitted their fraud prevention plans in May 2025 — a compliance milestone Taiwan's earlier, more ambitious platform law never managed to reach.
The Ghost of the DISA
To understand why June 2025 matters, consider June 2022. Three years earlier, Taiwan's National Communications Commission released a draft Digital Intermediary Services Act (DISA), a text explicitly modeled on the EU's Digital Services Act. It proposed annual risk assessments for platforms with more than 2.3 million Taiwanese users, algorithmic transparency disclosures, mandatory content moderation reporting, and fines of up to NT$10 million for non-compliance.
Within ten weeks, the NCC shelved the bill. Public opposition was fierce — not because Taiwanese internet users lack concern about platform accountability, but because Taiwan's historical experience with authoritarian speech control makes any government content-regulation authority feel different here than it does in Brussels. Civil society groups, opposition politicians, and platform operators raised overlapping concerns about selective enforcement risk and government overreach. A bill that reads as a democratic safeguard in European capitals read as a potential censorship toolkit to many Taiwanese critics. The DISA's collapse was not a failure of demand — it was a failure of framing.
The Anti-Fraud Reframe
The FCHPA succeeded where the DISA failed for one central reason: it anchored platform obligations in concrete, measurable harm. Fraud is not speech regulation. No reasonable observer disputes that removing advertisements designed to steal money from Taiwanese citizens is a legitimate state objective. The Act does not ask platforms to moderate political speech or assess "systemic risks" — a concept the DSA borrowed from financial regulation and applied to digital platforms with limited definitional clarity. It asks them to verify who is paying for ads and to remove content matching the statutory definition of fraud.
This distinction matters. The EU DSA's transparency framework is impressive in scope — very large online platforms must publish systemic risk assessments, content moderation statistics, algorithmic audits, and advertising repositories. But the DSA's breadth is also its enforcement vulnerability: when the legal obligation is to reduce "systemic societal risks," what counts as compliance becomes politically contested across twenty-seven member states. Taiwan's approach trades comprehensiveness for auditability. Fraud has a legal definition. A 24-hour removal window is measurable. Advertiser identity verification is checkable. The transparency reports platforms must file are specific enough that MODA can actually evaluate them.
What the Model Leaves Out — and What Fills the Gap
The strongest case for a comprehensive DSA-style law is worth stating plainly: fraud-only transparency cannot capture the full range of platform harms a democracy needs to address. The DSA's advertising repository, its notification requirements explaining why users see specific ads, and its vetted-researcher data access provisions address harms the FCHPA cannot touch — political micro-targeting, algorithmic amplification of health misinformation, cross-border influence operations. Taiwan, which faces sustained information operations from the People's Republic of China, has particular reason to care about influence-campaign transparency that no anti-fraud statute can mandate.
Taiwan's policymakers appear to know this. The NT$1.4 billion (approximately US$46.7 million) committed over three years for cybersecurity infrastructure — disclosed alongside the FCHPA transparency mandate — signals that MODA views the December 2025 reports as a starting point rather than a ceiling. Personal Data Protection Act amendments promulgated in November 2025 introduced a new independent Personal Data Protection Commission and 72-hour breach notification requirements aligned closer to GDPR standards, extending accountability into data governance territory the FCHPA does not reach. Taiwan's AI Basic Act, passed by the legislature in December 2025, further codified transparency and explainability as core governance principles across seven statutory pillars, suggesting Taipei is building accountability infrastructure layer by layer rather than attempting a single sweeping statute.
The Proportionality Case
What Taiwan demonstrates is that proportionality in platform regulation is not only about limiting government power — it is also about building compliance ecosystems that function. The DISA's collapse cost Taiwan roughly three years of platform accountability progress. The FCHPA's swift implementation cycle — law effective July 2024, compliance plans submitted May 2025, transparency reports due December 2025 — shows what targeted, harm-specific mandates can achieve when they do not threaten the speech rights of the regulated.
For non-EU jurisdictions watching the DSA and weighing whether to replicate it wholesale, Taiwan's experience offers a more tractable template: start with the clearest harms, build the compliance infrastructure, and expand scope from a position of demonstrated regulatory credibility rather than aspirational legislative ambition. The December 2025 reports, when they arrive, will be the first real test of whether that credibility has been earned — and whether Taiwan's sectoral model can scale to the influence-operation challenge that fraud law alone will never be enough to answer.