Taiwan

北京2027武統台灣？貼文悄改「這3字」被抓包 她嗆：死不認錯

The China Post | Local Language | News | Nov. 28, 2025 | Geopolitical Conflict and Disputes

President Lai Ching-te stated on November 26 that Beijing is targeting 2027 to prepare for an attack on Taiwan, prompting some media to report that China would invade Taiwan in 2027. However, the Presidential Office spokesperson Guo Ya-hui dismissed these reports as a "wrong and one-sided interpretation." Subsequently, Lai’s Facebook post was found to have quietly changed three characters, altering the phrase from "aiming to forcibly unify Taiwan" to "aiming to complete the preparations to forcibly unify Taiwan."

KMT legislator Wang Hongwei harshly criticized Lai for refusing to admit the alteration, calling such conduct unworthy of the presidency. Wang highlighted that Lai announced a significant national defense budget of 1.25 trillion from 2026 to 2033 while bluntly stating China's goal to complete forcible unification by 2027. Wang warned that such comments could damage Taiwan's international economic cooperation, investment, tourism, and heavily pressure the armed forces.

Wang also pointed out that the change to Lai’s wording only came after media scrutiny and public outcry, with the Presidential Office attempting to minimize the issue by labeling earlier reports as one-sided. Despite verbal denials of fault, the alteration on official platforms suggested recognition of the seriousness of the remarks. Wang condemned the president’s alleged rumor-spreading as irresponsible, potentially sacrificing public security for political purposes, and criticized the office’s deflection of responsibility onto the media.

11月消費者信心指數景氣指標創半年最高 學者提醒三大變數

United Daily News | Local Language | News | Nov. 28, 2025 | UndeterminedEconomic Growth

Central University's Taiwan Economic Development Research Center reported the November Consumer Confidence Index (CCI) at 64.65 points, a 0.69-point increase from October, marking a six-month high for five out of six key indicators. The five rising indicators include price levels, household economic conditions, domestic economic climate, job opportunities, and purchase of durable goods, with the latter showing the largest increase. However, consumer confidence in the timing to invest in stocks dropped to a five-month low at 27.02 points.

Executive director Wu Da-ren noted that strong export performance has boosted economic condition indicators and housing market confidence, which has rebounded for the first time since March. Despite rising economic indicators, the stock investment sentiment declined due to a significant recent stock market rally prompting profit-taking. Vehicle sales have recovered following tariff impacts, and recent central bank policy easing has increased bank lending capacity for home loans, contributing to improved housing market optimism.

Wu highlighted three major uncertainties for future economic trends: the impact of reciprocal tariffs between the U.S. and Taiwan, especially regarding Taiwanese investment flowing into the U.S. which could crowd out domestic private investment; the unclear effects of semiconductor tariffs on smaller Taiwanese supply chain companies; and the risk of an AI investment bubble triggered by recent losses incurred by major AI firms like OpenAI, whose uncertain profitability raises concerns about the sustainability of AI-driven economic growth in Taiwan.

大陸開源AI 下載量首超車美國

United Daily News | Local Language | News | Nov. 28, 2025 | UndeterminedTech Development/Adoption

Mainland China has overtaken the United States in the global market for open-source artificial intelligence (AI) models for the first time. Research by MIT and Hugging Face reveals that downloads of open AI models developed in mainland China now account for 17 percent of the global total, surpassing the 15.8 percent share held by U.S. developers. This growth is largely driven by Chinese companies such as DeepSeek and Alibaba’s Tongyi Qianwen.

Open models, which allow developers to freely download, modify, and integrate AI technologies, are seen as crucial for fostering innovation among startups and researchers. Mainland China's focus on open-source AI contrasts with the closed and proprietary strategies of leading U.S. tech firms like OpenAI, Google, and Anthropic, which control access to their advanced AI systems and monetize them through subscriptions and enterprise services.

Chinese companies have shifted toward open models, partly due to restrictions on advanced AI chip supplies from Nvidia and governmental encouragement. Analysts note that open source is more mainstream in China, whereas U.S. companies prioritize high valuations and keeping key technologies proprietary. The U.S. government, under the Trump administration’s AI Action Plan, has promoted open-model development reflecting "American values," but leading firms still limit openness. OpenAI’s recent release of open weight models is only partial, lacking full code and training data, while Meta is focusing more on closed AI models to compete in the race for superintelligence.