Loading your language..
Global AI safety strategy hinges on US-China collaboration, yet the two powers regard each other as the crux of the problem.

Global AI safety strategy hinges on US-China collaboration, yet the two powers regard each other as the crux of the problem.

C2🇹🇼 中文🇺🇸 English

September 17th, 2026

Global AI safety strategy hinges on US-China collaboration, yet the two powers regard each other as the crux of the problem.

C2
Please note: This article has been simplified for language learning purposes. Some context and nuance from the original text may have been modified or removed.

🇺🇸 English

As anxieties mount worldwide over the latent perils of artificial intelligence, the success of any global initiative hinges on wholehearted cooperation between the United States and Chinayet these two superpowers appear to be doing nothing but growing ever more suspicious of each other's AI strategies.

隨著世人對人工智慧所潛藏風險的憂慮與日俱增任何全球性方案能否竟其功端賴美中兩國的通力合作——然而這兩個超級大國似乎唯有對彼此的AI戰略愈發猜忌一途

AI governance is expected to feature on the agenda of next week's talks in Washington between President Trump and Chinese leader Xi Jinping.

AI治理預料將被納入下週於華盛頓舉行的川普總統中國領導人習近平會談的議程之中

The issue is becoming increasingly pressing in both countries, with American tech leaders saying that averting the worst-case scenarios that an unbridled AI might unleash hinges on a degree of cooperation with China.

此一議題在兩國日益迫在眉睫美國科技界領袖表示欲規避失控AI所可能引發的最壞情境端賴與中國一定程度的合作

Yet both nations are scrambling to secure a decisive advantage in the field of artificial intelligence, and Trump has accordingly warned that any regulatory measures targeting the technology would only benefit China.

然而兩國皆汲汲於爭奪人工智慧領域之決定性優勢川普遂警告任何針對該技術之監管舉措皆將助益中國

For this reason, any agreement reached at the forthcoming talks may prove merely superficial, yet it may nonetheless serve as the genesis of a grander design.

職是之故於即將舉行之會談中任何協議或僅具表面性質然亦不失為更宏大格局之濫觴

"A formal agreement or consensus may be well-nigh unattainable, yet they need not go that far," remarked Sam Sachs, a senior research fellow at the Future of American, Chinese, and Global Affairs Institute at Johns Hopkins University's School of Advanced International Studies.

正式協議或共識或許近乎遙不可及但他們無須走到那一步,」約翰霍普金斯大學高等國際研究學院美國、中國與全球事務未來研究所高級研究員薩姆·薩克斯表示

"Trump and Xi need only acknowledge that AI poses risks to both nations in order to carve out political space."

川普和習近平只需透過承認AI對兩國均構成風險便可開闢政治空間。」

Trump has dismissed calls to tighten regulation of the technology, asserting that doing so could allow China to overtake the United States in AI.

川普駁回了強化該技術監管籲求聲稱此舉恐致使中國在AI領域凌駕美國

He has repeatedly declared that the nation that holds sway over the technology "wins everything."

屢屢宣稱,在該技術上執牛耳之國家贏得一切」。

"We are leading China, and indeed every other nation," Trump declared on social media on Monday.

「我們正領先中國,乃至所有其他國家,」川普於週一在社群媒體上如是宣稱。

The two countries are advancing this technology along markedly divergent paths.

兩國正以迥異的路徑推進該項技術

American companies, drawing on the computing power driven by cutting-edge chips and the country's rapidly expanding data-centre infrastructure, are vying for the frontier of artificial intelligence; China, by contrast, is striving to promote the widespread global adoption of its open-source, more affordable models.

美國企業憑藉尖端晶片所驅動的運算能力以及該國急速擴張的資料中心基礎設施競逐人工智慧之前沿中國致力推動開源更為經濟實惠的模型於全球廣泛普及

Although US-led controls categorically prohibit China from obtaining the world's most advanced artificial-intelligence chips, Chinese AI models have nonetheless made considerable strides and won the favour of American companies.

儘管美國主導的管制措施嚴禁中國獲取全球最先進的人工智慧晶片中國的AI模型仍舊取得了長足進展博得美國企業的青睞

While most American frontier models are closed-source, their design elements shrouded in impenetrable secrecy, US Treasury Secretary Scott Bessent said on Tuesday that the United States urgently needs to develop more open-source models in order to counter China.

雖然大多數美國前沿模型皆屬閉源其設計元素諱莫如深,然而美國財政部長史考特·貝森特週二表示,美國亟需開發更多開源模型與中國相抗衡

The United States accused China of "aggressively and maliciously" attempting to extract capabilities from American AI systems, and China struck back, alleging that the United States aimed to suppress Chinese enterprises so as to establish an "AI industry monopoly."

美國譴責中國「積極、惡意」地試圖自美國AI系統中攫取能力,中國則予以反擊,指控美國意在打壓中國企業,從而確立「AI產業壟斷」。

David Sacks, Trump's chief AI adviser, asserted that China "has already refused" to slow the pace at which it is developing the most advanced forms of AI.

川普的首席AI顧問大衛·薩克斯聲稱中國已經拒絕放緩其開發最先進AI形式步伐

Speaking on Fox News's "The Ingraham Angle," he said that, in stark contrast to the nuclear tensions with the Soviet Union during the Cold War, it is quite simply impossible to verify whether China is abiding by any AI commitments.

他在福斯新聞的「英格拉罕角度」節目中表示,與冷戰時期和蘇聯之間核緊張局勢迥異要核實中國是否恪守任何AI承諾實屬不可能

"Under these circumstances, we have neither trust nor any means of verification," Sacks said.

在這種情況下我們既無信任可言亦無從核實,」薩克斯說

"Therefore, we must not foolishly hand over the entire race to China."

因此我們切莫愚不可及地將整場競賽拱手讓予中國。」

Nevertheless, individuals with close ties to the Chinese government continue to argue that coordination in the field of artificial intelligence is, in fact, a beneficial undertaking.

儘管如此,與中國政府過從甚密之人士仍主張,於人工智能領域展開協調,實屬裨益之舉。

At an annual security and defence conference held in Beijing, former Chinese ambassador to the United States Cui Tiankai remarked on Tuesday that AI is an area in which "we ought to advance and establish" dialogue between Beijing and Washington.

於北京召開的一場年度安全與國防會議上,前中國駐美大使崔天凱週二指出AI乃一我們理當推進並建立北京與華盛頓之間對話領域

"Because a need exists, and indeed a steadily growing need, cooperation is urgently required," he stated at the Xiangshan Forum.

緣於存在需求且係不斷增長之需求亟需合作,」他在香山論壇上如是表示

"So then, how are we to navigate this balance?"

那麼我們又當如何駕馭此一平衡?」

Two companies that hold sway over America's AI sectorAnthropic and OpenAIhave each imposed access restrictions on China, asserting that the United States, on national security grounds, has a necessary interest in preserving its leading position in AI technology.

兩家執美國AI領域牛耳之企業——Anthropic與OpenAI——分別對中國設下訪問限制並聲稱美國基於國家安全之考量有維護其在AI技術上領先地位之必要。

Yet both have also called for cooperation with China in order to manage the various problems that arise on the security front.

然二者亦呼籲與中國展開合作以管控安全層面之種種問題

Over the weekend, Dario Amodei, the chief executive of Anthropic, penned an article in which he called for a slowdown in AI development on safety grounds and argued that the United States ought to curb China's progress, including by maintaining the restrictions America has imposed on China's procurement of the world's most advanced AI chips.

Anthropic執行長達里奧·阿莫德伊於上週末撰文以安全為由籲請放緩AI發展並主張美國應遏抑中國之進展其途徑包括維持美國對中國採購全球最先進AI晶片所施加的限制

Nevertheless, he added, "Coordinating the pace on a global scale will inevitably require working hand in hand with China, and this authoritarian state possesses by far the most advanced AI capabilities to date."

然而,他亦補充道,「全球性的步調協調勢必需要與中國通力合作,而這個專制國家坐擁迄今為止最為先進的AI能力。」

His article promptly drew a sharp rebuke from the authorities in Beijing, with China's Ministry of Foreign Affairs stating on Monday, "Spreading panic, confrontation and vicious competition will only disrupt the global AI governance process and benefit no one."

他的文章旋即招致北京當局的嚴詞抨擊,中國外交部於週一表示,「散布恐慌、對抗和惡性競爭只會擾亂全球AI治理進程,對任何人都沒有好處。」

China's state-run newspaper Global Times went a step further, asserting that his remarks were "brimming with containment clauses aimed at China and essentially amount to a 'Cold War script' for the AI domain."

中國官方媒體《環球時報》更進一步他的言論充斥著針對中國的遏制條款本質上是一份AI領域的『冷戰劇本』。」

Trump had consulted Nvidia's chief executive, Jensen Huang, and even phoned the world's most advanced computer chip manufacturer midway through a podcast interview on Monday.

川普曾諮詢輝達執行長黃仁勳甚而於週一播客採訪進行中途致電這家全球最先進電腦晶片之製造商

Nvidia broadly supports exporting some chips to China, thereby building an American ecosystem around the technology and fostering a more open relationship in AI, given that China possesses a great many researchers.

輝達大體上支持向中國出口部分晶片藉此為該技術構築美國生態系統,並在AI方面建立更為開放之關係蓋因中國擁有眾多研究人員

In both the United States and China, the ever-increasing capabilities of AI have left senior officials and policymakers profoundly alarmed.

在美國與中國,AI日臻強大的能力,皆令高層官員及政策制定者為之震懾。

In an article published on Sunday, the head of China's Ministry of State Securitythat is, its spy agencycautioned that artificial intelligence poses a multitude of threats to the nation, spanning from political and ideological security to cyberattacks.

中國國家安全部——亦即其間諜機構——負責人於週日刊發的一篇文章中告誡,人工智慧對國家構成諸多威脅,其範疇橫跨政治與意識形態安全,乃至網路攻擊。

In that article, Chen Yixin argued that artificial intelligence, if manipulated by those with "ulterior motives," could imperil China's political and ideological security, and he specifically named the powerful AI models developed by Anthropic and OpenAI.

在該文中陳一新主張人工智慧倘若別有用心之徒操弄危及中國政治意識形態安全,並特別點名AnthropicOpenAI所研發強大AI模型

The article asserted that such models might be used to "weaponize" cybersecurity vulnerabilities, thereby threatening China's critical information infrastructure.

文章此等模型或可被用以網路安全漏洞武器化」,進而威脅中國關鍵資訊基礎設施

Paul Triolo, a partner at the DGA-Albright Stonebridge Group, pointed out that there have recently been several AI-related incidents that have given rise to misgivings within China regarding AI governance.

DGA-Albright Stonebridge Group合夥人保羅·特里奧洛指出近期尚有數起與AI相關之事件於中國境內引發了對AI治理的疑慮。

One such instance was a US security firm's recent detection of an AI vulnerability in WeChat, which, as China's principal instant-messaging application, also incorporates payment and numerous other functions.

其中一例一家美國安全公司近日於微信中偵測到AI漏洞;微信作為中國主要之即時通訊應用亦兼具支付等多項功能

Although Chinese officials asserted, following the conclusion of the Beijing summit between Trump and Xi in mid-May, that the two countries had already agreed to initiate a dialogue on the development and governance of artificial intelligence, tangible progress on the matter has thus far remained conspicuously scant.

儘管中國官員川普與習近平五月中旬北京峰會落幕後聲稱中美兩國業已同意就人工智慧的發展與治理展開對話然相關進展迄今仍寥寥可數。

Meanwhile, China and the United States have continued to form artificial intelligence alliances with like-minded countries.

與此同時,中美兩國持續與理念相契的國家籌組人工智能聯盟。

In July, China announced the establishment of the World Artificial Intelligence Cooperation Organization, whose founding member states numbered twenty-nine, with both Russia and Pakistan among them, the organization's purpose being to promote global AI governance.

於七月中國宣告成立世界人工智慧合作組織其創始成員國計二十九俄羅斯與巴基斯坦皆列其中,該組織之宗旨在於促進全球AI治理

In the same month, China's Xi Jinping urged the member states of BRICS to accelerate the construction of a global AI governance framework and to establish an open-source AI community.

同月中國之習近平敦促金磚國家成員國加速建構全球AI治理框架並創設開源AI社群

Some political observers interpret this China-led organization as a response to the US-led "Silicon Peace Initiative"; that initiative was launched last year, and its central aim is to strengthen collaboration with countries such as Japan, the United Kingdom, and Australia in the artificial intelligence supply chain.

若干政治觀察家將此一由中國主導之組織詮釋為對美國所主導之「矽和平倡議」的回應該倡議於去年啟動其要旨在于強化與日本、英國及澳洲等國在人工智慧供應鏈上協作

In an exclusive interview on CBS's "Face the Nation," when asked about his expectations for the Trump-Xi meeting regarding AI, Amodei remarked, "The more enduring imperative will be for the two sides to work in concert to set a ceiling on the pace of AI's advancement."

於接受CBS「面對全國」節目專訪、被詢及對川習會就AI所抱持之期望時阿莫德伊表示:「更為長遠之要務將是雙方協同致力為AI進展之速度劃定上限。」

He candidly acknowledged, however, that such an undertaking would likely prove exceedingly arduous.

惟其坦言此舉恐極為艱難

On the eve of the Trump-Xi summit, the account "Yuyuan Tantian," which is closely associated with Chinese state media, likewise adopted a cautious stance in its recent social media posts.

值川習會前夕與中國官方媒體過從甚密之帳號玉淵譚天」,其邇來於社群媒體所發布之貼文亦復秉持審慎立場

"Only if Washington takes the lead in demonstrating that its safety regulations apply equally to its own model companies will the two sides, the United States and China, have any possibility of embarking on substantive consultations," the post read.

「唯有華盛頓率先證明其安全規範對本國模型企業一體適用,美中雙方方有可能展開實質性磋商,」該貼文寫道。

September 17th, 2026

Trending Articles

King and AI: Charles Meets AI Leaders as Safety Concerns Persist

King and AI: Charles Meets AI Leaders as Safety Concerns Persist

國王與人工智慧:查爾斯會晤人工智慧領袖,安全隱憂持續蔓延

C2Sep 18
With his tenure drawing to a close and global crises looming on every side, the UN Secretary-General discusses the road ahead.

With his tenure drawing to a close and global crises looming on every side, the UN Secretary-General discusses the road ahead.

卸任在即,全球危機四伏,聯合國秘書長論前路

C2Sep 18
Democratic presidential candidates vie to respond to AI threats; Trump remains indifferent.

Democratic presidential candidates vie to respond to AI threats; Trump remains indifferent.

民主黨總統候選人競相回應AI威脅 特朗普淡然處之

C2Sep 18
House passes bill aimed at addressing data centers' impact on energy costs

House passes bill aimed at addressing data centers' impact on energy costs

眾議院通過法案 旨在應對資料中心對能源成本的影響

C2Sep 18
OpenAI exposes concerning new AI behavior and pledges to enhance tracking.

OpenAI exposes concerning new AI behavior and pledges to enhance tracking.

OpenAI 揭露令人憂慮的新 AI 行為,並承諾加強追蹤

C2Sep 17
Huawei Unveils New Chip Technology as Chinese Firms Intensify AI Race with Nvidia

Huawei Unveils New Chip Technology as Chinese Firms Intensify AI Race with Nvidia

華為發布新晶片技術 中企加緊與輝達展開AI競賽

C2Sep 17
The tech industry is divided over calls to slow down AI coordination.

The tech industry is divided over calls to slow down AI coordination.

科技業對AI減緩協調呼聲現分歧

C2Sep 17
Trump downplayed the necessity of checks on AI development, saying he was unwilling to cede the advantage to China.

Trump downplayed the necessity of checks on AI development, saying he was unwilling to cede the advantage to China.

川普淡化AI發展檢查之必要,並稱不願讓優勢拱手於中國

C2Sep 14
Oprah Winfrey invites you to find your "AHA" moment at the Sphere

Oprah Winfrey invites you to find your "AHA" moment at the Sphere

歐普拉·溫芙蕾邀你在Sphere尋覓你的「AHA」時刻

C2Sep 14