A renewed push, an old debate
On May 31, 2026, President Lai Ching-te met a group of high-school students and came away endorsing tougher action against Chinese information operations, calling the influence of apps such as TikTok and RedNote a potential "national security issue" and warning that, mishandled, "Taiwan could lose its will to safeguard democracy over time" (Taipei Times). With local elections due in November 2026 — the next major test of Beijing's "cognitive warfare" — the comments reopen a question Taiwan has wrestled with for four years: how do you blunt platform-borne disinformation without handing the state a lever over speech?
Taiwan's answer, so far, is unusual among democracies. It has chosen targeted, sector-specific tools over a single omnibus platform law — and that choice deserves more attention than it gets.
What Taiwan actually built
The scale of the problem is not in dispute. Taiwan's National Security Bureau reported that in 2025 it identified more than 2.3 million pieces of disinformation and over 45,000 fake or bot accounts tied to PRC cognitive-warfare operations run through Chinese IT and marketing firms (NSB report, via OCAC). Rather than meet that with a DSA-style intermediary statute, Taipei assembled a patchwork:
- Anti-fraud platform rules. The Fraud Crime Hazard Prevention Act took effect on July 31, 2024, and four subsidiary regulations under the Ministry of Digital Affairs (MODA) came into force in late 2024. Large advertising platforms — Google, YouTube, Line, Facebook, Instagram and TikTok — must verify advertiser identities, file fraud-prevention plans, publish transparency reports, and remove flagged fraudulent ads within 24 hours, on pain of fines up to NT$10 million (MODA).
- Targeted app bans. In December 2025 the government imposed a one-year ban on RedNote (Xiaohongshu), citing fraud risk and data security after the app had passed three million Taiwanese users (SCMP). TikTok and its sister app Douyin have been barred from public-sector devices since 2022.
- A voluntary transparency code. In January 2026 the NCC unveiled two non-binding guidelines covering YouTube, Facebook, Line, TikTok and Dcard. Platforms are asked to disclose content-moderation criteria, offer Traditional-Chinese complaint channels and publish transparency reports; the NCC will grade them by end-2026 — but, as its own secretary-general conceded, "platforms would not be penalized if they do not follow them" (Taipei Times).
Why not a DSA?
The strongest case for a binding law is real. Voluntary codes have no teeth; a hostile state actor flooding the zone with millions of falsehoods a year is not deterred by a politely worded transparency guideline, and grading platforms after the November vote does nothing to protect that vote. Comparable democracies — the EU, the UK, Australia — have concluded that only enforceable obligations move platforms the size of Google and Meta.
Taiwan tried that path and abandoned it. The NCC's draft Digital Intermediary Services Act, released in June 2022, would have let agencies flag content and seek court-backed "information restraint" orders. Within weeks more than 30,000 citizens objected on the government's public-consultation platform against fewer than 150 in support, and by September 8, 2022 the NCC halted drafting, conceding it was back to "square one" (Taipei Times). The backlash was bipartisan and rooted in a precise fear: a government empowered to flag "disinformation" is a government that can flag dissent — especially dangerous in a society Beijing is actively trying to divide.
The case for the patchwork
That collapse looks less like failure and more like institutional learning. Taiwan's current toolkit is narrower, and narrowness is a feature.
The anti-fraud rules regulate conduct — identity verification, ad transparency, scam takedowns — not viewpoints. They target the commercial-fraud vector that overlaps heavily with influence operations, without asking a regulator to adjudicate what is true. The app bans are blunt but legible: they restrict specific foreign-controlled platforms on stated security grounds, a decision voters can debate and a future government can reverse, rather than a standing censorship power. And the voluntary NCC code creates disclosure pressure and public benchmarking without criminalizing speech or deputizing platforms as state censors.
This is close to the proportionate, evidence-based model we favor: address the demonstrable harm — fraud, coordinated inauthentic behavior, foreign data risk — with the least speech-restrictive instrument that fits, and keep contested judgments about "truth" out of government hands. It also leans on Taiwan's real comparative advantage: a dense civil-society fact-checking ecosystem and the media-literacy and history education Lai again emphasized on May 31.
The approach is not costless. Voluntary codes may prove too weak if platforms simply ignore them, and app bans invite tit-for-tat retaliation and set precedents that less liberal governments will happily cite. The honest response is to keep the bans evidence-based and time-limited, publish the NCC scorecards prominently enough that disclosure actually bites, and resist the temptation — likely to resurface after November — to revive an omnibus speech law. Taiwan has shown that a democracy under sustained information assault can defend itself without becoming what it is defending against. That lesson is worth exporting.