Taiwan has developed one of Asia's most pragmatic — and constrained — approaches to platform disinformation regulation: repurposing a narrowly scoped anti-fraud law to fill the hole left by a failed comprehensive platform bill. The arrangement is producing real enforcement results against global platforms. It is also proving inadequate for the disinformation threats Taiwan faces most acutely.
The Bill That Died in Two Months
The story begins in June 2022, when Taiwan's National Communications Commission (NCC) proposed the Digital Intermediary Services Act (DISA), modelled on the European Union's Digital Services Act. The bill would have imposed transparency obligations, content removal duties, and algorithm accountability requirements on large platforms. Within weeks, it was dead. More than 30,000 people voted against it on the government's public participation platform; fewer than 150 expressed support. Civil society groups, industry, and opposition legislators accused the NCC of seeking sweeping executive discretion over digital speech — invoking uncomfortable memories of Taiwan's authoritarian past. The bill was shelved before reaching the Legislative Yuan in August 2022.
That public verdict was legitimate. Poorly scoped platform legislation that grants regulators broad content-removal powers without independent oversight genuinely does pose speech risks. The criticism was not paranoia; it reflected hard lessons from governments that have weaponised disinformation laws against political opponents. Taiwan's democratic instincts were working as intended.
Anti-Fraud Law Steps In
Rather than revive the DISA, Taiwan pivoted to sector-specific amendments to existing laws. The most consequential result was the Fraud Crime Prevention Act, which took effect on July 31, 2024. Although designed primarily to combat online financial fraud, the law contains provisions that function as a platform accountability mechanism for a subset of disinformation: commercially deceptive content.
Under the Act, large online advertising platforms must disclose the identities of advertisers and their funding sources, and must promptly remove fraudulent ads upon notification by the competent authority. Article 42 further authorises the Ministry of Interior to impose DNS blocking against platforms that pose an emergency fraud risk.
That last power was used in December 2025, when the Ministry of Interior ordered a one-year block against Xiaohongshu (RedNote), the Chinese social media platform, citing fraud losses and invoking the emergency provision. The numbers behind the action were real: cumulative fraud losses linked to Xiaohongshu in Taiwan had reached NT$132.9 million (~US$4.1 million), with NT$114.77 million of that occurring in the first eleven months of 2025 alone. Over three million Taiwanese users were on the platform at the time of the block.
Critics argued the ban was disproportionate and geopolitically convenient — Xiaohongshu's Chinese ownership made it an easy target when the fraud-prevention rationale could have been addressed through less drastic content-level enforcement. The concern is fair. Using an anti-fraud statute's emergency clause to remove an entire platform serving three million users is a significant step, and the lack of a transparent adjudicatory process is a structural weakness in this legal framework.
Enforcement Against Global Platforms
The more durable — and proportionate — use of the Fraud Crime Prevention Act has been its application to Meta. In May 2025, Taiwan's Ministry of Digital Affairs (MODA) issued its first fine under the law: NT$1 million (~US$33,000) against Meta for failing to disclose the commissioners and funding sources of two Facebook advertisements. It was a landmark action, establishing that the law would be applied to dominant global platforms, not only smaller operators.
Enforcement escalated from there. By October 2025, total fines against Meta had reached NT$18.5 million (~US$600,000), including a NT$2.5 million penalty for failing to remove false advertisements after an official government order. A recurring friction point was procedural: Meta cited difficulties processing paper-format takedown requests, leading to URL transcription errors that delayed removal. MODA responded by launching a digital fraud reporting platform, enabling county and city governments to submit URLs electronically, and three municipalities had adopted the system by late 2025.
These are not trivial enforcement actions. They demonstrate that a targeted statutory framework, consistently applied, can impose real compliance pressure on global platforms within a democracy that lacks a comprehensive DSA-equivalent.
Where the Patchwork Breaks Down
The anti-fraud framing captures content that is deceptive and commercially transactional. It does not reach state-sponsored influence operations, coordinated inauthentic behaviour, or AI-generated political disinformation — which are precisely the threats Taiwan faces most intensely.
During the January 2024 presidential election, AI-generated deepfake videos attempted to discredit candidates, and foreign information operations amplified pro-China narratives through bot networks. Taiwan's legislature responded in 2023 by amending the Public Officials Election and Recall Act and the President and Vice President Election and Recall Act to increase penalties for deepfake candidate disinformation. But prosecution requires attributing specific false content to identifiable individuals — a near-impossible standard when the actors are foreign-controlled bot networks operating across jurisdictions.
Taiwan's AI Basic Act, promulgated in January 2026, establishes risk-classification principles and transparency requirements for AI systems but does not create platform-specific disinformation obligations. It provides a governance skeleton without the muscle of enforcement.
Civil society fills some of this gap. The Taiwan FactCheck Center, Cofacts, Doublethink Lab, and several citizen-operated fact-checking services have built a functioning civic infrastructure for disinformation detection. The government broadly supports these efforts, and they matter. But civic constitutionalism and statutory regulation are complements, not substitutes. Fact-checkers cannot compel platforms to demonetise state-sponsored influence operations.
The Proportionality Argument
Taiwan's predicament is instructive for other democracies weighing similar trade-offs. The DISA's 2022 failure was a democratic verdict against executive overreach — not a permanent veto on platform accountability legislation. A framework designed with meaningful parliamentary oversight, independent adjudication of removal orders, and transparent appeals processes would address the legitimate objections that sank the DISA. The EU's DSA, imperfect as it is, provides a reference architecture from which Taiwan could draw selectively.
The current approach — anti-fraud enforcement plus civil society plus election law amendments — is getting results within a narrow perimeter. Expanding that perimeter without replicating the DISA's structural flaws is Taiwan's next legislative challenge.