當每個台灣人都成為AI標籤
2300萬筆戶籍記錄、170名政治人物、75個政黨、24000個公民組織——這些台灣社會的基本組成,已經被轉化為中國國企GoLaxy系統中的數據點。每個條目都附帶「對中態度」標籤,成為AI自動生成認知戰內容的原料。2024年總統大選期間,這套系統已經完成實戰測試。2026年地方選舉,規模更大的操作正在準備中。
這不是科幻小說的情節。根據GoLaxy系統的技術架構分析,這套工具已經超越傳統的資訊監控範疇,建構出完整的「認知操控基礎設施」。系統能夠根據目標受眾的特徵,自動產生客製化的宣傳內容——針對特定選區、特定議題、特定社群,投放經過演算法優化的訊息。
AI實力與防禦能力的矛盾落差
諷刺的是,台灣本身正是全球AI產業鏈的關鍵樞紐。台積電(TSMC)為NVIDIA、Google、Apple代工AI晶片,沒有台灣的半導體製造能力,全球AI革命將陷入停滯。2017年行政院提出「AI小國大戰略」,Taiwan AI Academy培養出數千名AI人才。台灣在AI硬體與人才培育上的實力,與認知戰防禦能力之間,存在巨大落差。
這個落差源自對威脅本質的誤判。GoLaxy系統針對的不是技術系統,而是人類認知系統——記憶、情感、身分認同。當AI能夠分析24000個公民組織的政治傾向,自動生成針對環保團體、勞工組織、宗教社群的差異化訊息時,攻擊的對象是台灣社會的集體意識結構本身。
從新聞自由到數位主權的新戰場
台灣在2024年無國界記者組織(RSF)世界新聞自由度指數中排名第35位,位居亞洲前列。從1993年有線電視法終結三台壟斷,誕生24小時新聞台,到解嚴後媒體生態從少數黨營報紙擴張至數百家平台,台灣建立起相對多元的媒體環境。
但新聞自由保障的是言論市場的開放競爭,數位主權面對的是演算法控制的資訊戰。當GoLaxy系統能夠追蹤2300萬人的數位足跡,分析其政治傾向,再透過社群平台演算法精準投放客製化內容時,傳統的新聞自由框架已經不足以應對。問題不在於是否有審查,而在於誰掌握了定義「你會看到什麼」的演算法權力。
建構防禦的三個方向
面對自動化認知戰,防禦必須建立在三個層次。第一是數位素養教育:公民需要理解演算法如何運作、如何辨識客製化宣傳、如何追溯資訊來源。這不是技術問題,而是新時代的公民教育核心。
第二是平台透明度要求。社群媒體必須揭露內容推薦機制、廣告投放者身分、跨境資訊流動路徑。台灣在2026年選舉前,需要建立法律框架,要求平台公開演算法決策邏輯,特別是涉及政治內容的推薦機制。
第三是跨國民主聯盟。GoLaxy系統針對的不只是台灣,所有民主體制都是目標。與立陶宛、捷克、日本等同樣面臨認知戰威脅的國家,分享威脅情資、技術對策、法律經驗,建立民主國家的集體防禦網絡。
當2300萬台灣人的資料已經被轉化為AI認知戰的彈藥庫,反擊的方式不是建立防火牆隔絕資訊,而是建構公民的認知免疫系統——理解威脅的運作機制,辨識操控的手法,保護集體記憶與身分認同不被演算法重新定義。這是台灣AI實力必須轉向內部防禦的時刻。
— 施郁雯
23M Taiwanese Profiled in Beijing’s AI System: GoLaxy and Automated Cognitive Warfare
When Every Taiwanese Becomes an AI Label
Twenty-three million household records, 170 political figures, 75 political parties, and 24,000 civil society organizations—the fundamental components of Taiwanese society have been converted into data points within GoLaxy, a system operated by a Chinese state-owned enterprise. Each entry carries an “attitude toward China” label, serving as raw material for AI-generated cognitive warfare content. During the 2024 presidential election, this system completed its live combat test. For the 2026 local elections, operations on a larger scale are already in preparation.
This is not science fiction. According to technical analysis of GoLaxy’s architecture, the tool has transcended traditional information surveillance, constructing a complete “cognitive manipulation infrastructure.” The system can automatically generate customized propaganda content based on target audience characteristics—tailored to specific electoral districts, specific issues, and specific communities, delivering algorithmically optimized messages.
The Paradox Between AI Strength and Defense Capability
Ironically, Taiwan itself stands as a critical hub in the global AI supply chain. TSMC manufactures AI chips for NVIDIA, Google, and Apple; without Taiwan’s semiconductor manufacturing capabilities, the global AI revolution would grind to a halt. In 2017, Taiwan’s Executive Yuan launched the “AI Small Nation, Big Strategy” initiative, and Taiwan AI Academy has trained thousands of AI professionals. Yet a vast gap exists between Taiwan’s prowess in AI hardware and talent cultivation and its cognitive warfare defense capabilities.
This gap stems from a misidentification of the threat’s nature. GoLaxy targets not technical systems but human cognitive systems—memory, emotion, and identity. When AI can analyze the political leanings of 24,000 civil organizations and automatically generate differentiated messages targeting environmental groups, labor organizations, and religious communities, the attack targets Taiwan’s collective consciousness structure itself.
From Press Freedom to the New Battlefield of Digital Sovereignty
Taiwan ranked 35th in Reporters Without Borders’ (RSF) 2024 World Press Freedom Index, among the top in Asia. From the 1993 Cable Television Act that ended the three-network monopoly and birthed 24-hour news channels, to the post-martial law media ecology expanding from a few party-run newspapers to hundreds of platforms, Taiwan has built a relatively pluralistic media environment.
But press freedom guarantees open competition in the marketplace of ideas; digital sovereignty confronts information warfare controlled by algorithms. When GoLaxy can track the digital footprints of 23 million people, analyze their political leanings, and precisely deliver customized content through social media algorithms, traditional press freedom frameworks prove insufficient. The issue is not whether censorship exists, but who controls the algorithmic power to define “what you will see.”
Three Directions for Building Defense
Facing automated cognitive warfare, defense must be built on three levels. First is digital literacy education: citizens need to understand how algorithms work, how to identify customized propaganda, and how to trace information sources. This is not a technical issue but the core of civic education for a new era.
Second is platform transparency requirements. Social media must disclose content recommendation mechanisms, advertiser identities, and cross-border information flow pathways. Before the 2026 elections, Taiwan needs to establish a legal framework requiring platforms to reveal algorithmic decision-making logic, especially for political content recommendation mechanisms.
Third is a transnational democratic alliance. GoLaxy targets not only Taiwan; all democratic systems are potential targets. Sharing threat intelligence, technical countermeasures, and legal experiences with countries like Lithuania, Czech Republic, and Japan that face similar cognitive warfare threats can build a collective defense network for democracies.
When data on 23 million Taiwanese has been converted into ammunition for AI cognitive warfare, the response is not building firewalls to isolate information, but constructing citizens’ cognitive immune systems—understanding threat mechanisms, identifying manipulation tactics, and protecting collective memory and identity from algorithmic redefinition. This is the moment Taiwan’s AI capabilities must turn toward internal defense.
— Curator