臺灣如何在2024年總統大選中揭穿錯假資訊？3個事實查核機構的案例分析與比較

Abstract

臺灣在全球錯假資訊研究中具有指標性地位，過去二十五年以來，多元民主中心（V-Dem）的資料顯示，臺灣始終是受到境外錯假資訊攻擊最嚴重的國家之一，臺灣亦被視為是錯假資訊戰的預警指標，尤其是在 2024 年這個全球超級選舉年。

當事實查核成為應對此危機的重要手段，查核組織試圖透過強調公正性與專業標準來建立正當性。但此一追求亦可能限制其回應其他新聞價值的能力，例如避免落入「假中立」（false equivalence），此外，查核組織也經常被質疑具有政黨偏好。

儘管國際學術研究已廣泛探討事實查核的知識論基礎、制度根源、制度化、組織與科技平台合作所帶來的影響，臺灣的經驗仍相對缺乏深入研究。事實查核常被視為一項源自美國並擴散至全球的實踐，然而，台灣的發展路徑呈現出不同於既有模式的特徵，進一步豐富並挑戰既有理論框架。

為補足上述研究缺口，並檢視台灣錯假資訊與事實查核的動態，本研究透過比較內容分析法，檢視在臺灣 2024 年總統大選期間，三個事實查核組織的查核報告：台灣事實查核中心、麥擱騙（MyGoPen）與亞洲事實查核實驗室。研究首先從組織間的共通性出發，歸納整體趨勢，包括錯假資訊的主題、正副總統候選人提及、內容形式、欺騙手法，以及組織所使用的參考資料與查核取向，並進一步比較三個組織在多個關鍵面向上的差異。

研究結果顯示：(1) 國際關係與衝突是查核報告中最主要的議題（24.25%）；(2) 涉及正副總統候選人的查核報告比例偏低，且在不同組候選人之間，分布相對平均（介於4.65% 至 7.97% 之間）；(3) 儘管錯假訊息越來越常透過影片等視覺形式呈現，文字仍是資訊傳播的核心元素；(4) 錯假資訊使用的欺騙技巧，絕大部分（90.70%）為虛構捏造與扭曲誤導；(5) 查核組織主要使用政府、媒體與開源情資進行查核，且 85.71% 的查核項目引用了多種類型的參考資料，顯示出資料的豐富與平衡；(6)查核工作的取向，主要關注網路瘋傳的錯假訊息（84.72%），針對公眾人物或機構言論的查核報告則顯著較少（15.28%）。

三家事實查核組織在議題選擇與查核取向上呈現出顯著差異。AFCL 著重於政治事實查核（64.44%）與國際事務（48.89%）；MyGoPen 則以查核網路瘋傳內容（98.33%）及公共衛生與醫療議題（35.00%）為主；相較之下，TFC 介於兩者之間，主要在查核網路瘋傳內容（88.97%），但是在政治事實查核上的比重高於 MyGoPen，且對內政議題的關注程度亦較高（27.94%）。這些不同的編輯選擇，可能源於組織在制度背景、組織資源及資金結構的差異。

藉由採取比較與數據導向的研究設計，本研究旨在呈現臺灣在關鍵情境下，錯假資訊生態的整體樣貌，也進一步說明各事實查核組織如何以不同方式理解並處理錯誤資訊。

Taiwan occupies a pivotal position in global disinformation research; for over twenty-five years, V-Dem data have ranked Taiwan among the most heavily targeted countries for foreign disinformation, and it is widely described as a “canary for disinformation,” particularly during the 2024 global election cycle. As fact-checking has become a response to this challenge, it has sought legitimacy through claims of impartiality and professional standards. Yet impartiality is not without tension: it may conflict with the need to avoid false equivalence, and fact-checkers frequently face accusations of partisanship.

While international scholarship has examined fact-checking’s epistemology, institutional roots, and the effects of institutionalization and platform partnerships, Taiwan’s experience remains underexamined. Although often framed as a U.S.-originated global practice, fact-checking in Taiwan reflects a distinct developmental trajectory that complicates established theoretical frameworks.

To address this gap and examine false information and fact-checking dynamics in Taiwan, this research adapts comparative content analysis to examine fact-checking reports from three organizations—Taiwan FactCheck Center (TFC), MyGoPen, and Asia Fact Check Lab (AFCL)—during Taiwan’s 2024 presidential election campaign. The study identifies both shared patterns in false information, including topics, mentions of presidential tickets, content formats, deception techniques, as well as the reference sources and fact-checking approaches used by these organizations while comparing differences across key dimensions.

Findings indicate that international relations were the most prevalent topic (24.25%), while fact-checking related to the three presidential tickets was relatively limited and closely distribute, ranging from 4.65% to 7.97%. Although disinformation increasingly utilized visual formats like video, text remained central to dissemination. The overwhelming majority of deception techniques involved misleading and fabricated content (90.70%). Fact-checkers primarily relied on official sources, professional media, and OSINT, and 85.71% of cases cited multiple source categories, showing strong source diversity. Most efforts targeted viral online false information (84.72%), with significantly fewer cases addressing claims by public figures and institutions (15.28%).

Notable differences emerged across the three organizations mainly in topics and fact-checking approaches. AFCL emphasized political fact-checking (64.44%) and international affairs (48.89%), while MyGoPen prioritized debunking viral content (98.33%) and public health and healthcare issues (35.00%). TFC occupied an intermediate position, focusing primarily on viral content (88.97%) but devoting a larger share of its work to political discourse than MyGoPen and placing greater emphasis on internal affairs (27.94%). These divergent editorial priorities may result from variations in their institutional origins, organizational capacities, and funding structures.

By adopting a comparative, data-driven design, this study captures both the overall contours of Taiwan’s false information environment and the distinct ways individual organizations interpret and address false information during high-stakes moments.