Taiwan's Digital Ministry is reviewing how artificial intelligence is labeled for the public after U.S. President Donald Trump posted on Truth Social on Sept. 25 that the field should be called 'super intelligence.' The review, run with the National Development Council and the Financial Supervisory Commission, is focused on consumer-facing terminology in Chinese and English. It is not a new safety test, a certification, or a model evaluation.
The distinction matters. Trump's wording was a comment, not a regulatory order. The post came as Washington and Beijing wrapped a summit in which Taiwan was not a central public issue; the State Department said its Taiwan policy remained unchanged. But it arrived while Taiwan's agencies are already trying to standardize how they describe AI services. Consumers see 'AI' on banking apps, government chatbots, health portals, and retail recommendation engines. The question for the Digital Ministry is whether adding 'super intelligence' to that vocabulary would help people understand what they are using—or make the technology sound more capable than anyone has verified.
What the review covers
According to the scope set for the review, the Digital Ministry is looking at public-facing labels. The NDC is examining how government agencies describe AI in official communications. The FSC is looking at financial-sector disclosures and chatbot interfaces. Neither agency has proposed a legal definition of 'super intelligence.' Instead, the exercise is about clarity: when a service is called AI, what does that mean for the person on the other side of the screen?
What is being evaluated? Terminology, not performance. The review does not appear to include benchmark results, bias audits, or accuracy rates. By whom? A cross-agency working group, with sector regulators retaining their existing authority. The public cannot yet check a list of approved terms, because no such list has been published.
That leaves a gap between a label and a guarantee. A bank can say its chatbot uses AI. A developer can call a model 'super intelligence.' Under Taiwan's current rules, neither phrase automatically triggers a higher level of scrutiny. The Consumer Protection Act and the Financial Consumer Protection Act require claims to be accurate and not misleading. They do not require a specific AI label. If a company markets a product as super intelligence, it still has to substantiate what it promises under existing advertising and consumer rules.
What Taiwan requires this month
As of late September, no Taiwan law or regulation requires the term 'super intelligence.' No agency has defined the threshold for using it. The FSC has separate guidance for banks that deploy chatbots, but that guidance is about disclosure and accountability, not about proving a system is super intelligent. The Digital Ministry's review is framed as a terminology exercise, which suggests recommendations rather than binding rules. That means the enforceable standard this month remains general: do not deceive consumers.
The review also has to answer a practical question. If 'super intelligence' is used in a government pamphlet, a procurement document, or a school lesson, what does it describe? Taiwan's existing terms are already broad. 'AI' covers everything from translation tools to large language models. 'Generative AI' is narrower but still includes systems with very different risk profiles. 'Super intelligence' would be broader still, unless regulators define it as a marketing term rather than a technical category.
Industry may move faster than the review. Taiwanese AI developers could adopt the phrase for investor decks or product launches, especially if it travels from U.S. social media. Enterprise buyers are more likely to ask about latency, cost, data residency, and integration. Government agencies tend to prefer established terms because procurement rules favor clear specifications. A new label could create confusion in contracts if it has no agreed meaning.
Other regulators offer no easy template. The European Union's AI Act sorts systems by risk, not by intelligence. The United States has no single naming rule. Taiwan's approach is sectoral: the FSC covers finance, the Ministry of Health and Welfare covers health, and the Digital Ministry coordinates digital policy. A 'super intelligence' label would not automatically trigger higher scrutiny in any of those sectors.
Taiwan's recent tech rulemaking shows why implementation matters. The satellite communications articles in the Telecommunications Management Act were passed in August but remain not in force while NCC procedures are pending. A terminology review can follow a similar path: an announcement, a draft, a consultation, and then a long wait before anything changes in practice. For now, the safe reading is that a review is underway, not that a new rule exists.
What to watch
Three signals will show whether the review changes anything. First, whether the Digital Ministry publishes a glossary or style guide for official AI language. Second, whether the FSC updates its chatbot guidance to address new labels. Third, whether any agency ties a term to a test, audit, or registration requirement. Without one of those steps, 'super intelligence' remains a phrase, not a regulatory category.
Consumers should treat it that way. A label can tell you what a company wants you to think. It does not tell you what was evaluated, by whom, or against what standard. Taiwan's review may clarify that boundary. Trump's wording may shape headlines, but the local question is narrower: when a public-facing AI service uses a new name, what can a Taiwanese consumer actually check?