Loading your language..
The divergence within the tech industry over slowing the pace of AI development is increasingly widening.

The divergence within the tech industry over slowing the pace of AI development is increasingly widening.

C1🇹🇼 中文🇺🇸 English

September 17th, 2026

The divergence within the tech industry over slowing the pace of AI development is increasingly widening.

C1
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

Concerns about artificial intelligence spiraling out of control, along with proposals to slow down the technology's development, are nothing new, but as warnings about chaos caused by runaway AI systems grow louder, these concerns and proposals have recently attracted renewed attention.

對於人工智慧失控的擔憂以及放緩該技術發展的建議並非新鮮事隨著有關失控AI系統造成混亂的警告日益高漲這些擔憂與建議近來再度受到關注。

Senior executives at several leading AI companies have recently spoken in unison about the need for coordinated mitigation measures, including international cooperation, to safeguard humanity from the most severe consequences.

數位領先的AI企業高階主管近期一致指出有必要推動協調性的減緩措施包括國際合作以保障人類免於最嚴峻的後果

However, opinions diverge, and some technology leaders are deeply unhappy with the claim that their companies cannot ensure safety on their own.

然而各方看法分歧部分科技界領袖對於這樣的主張深感不滿其公司無法獨自確保安全

The following is a summary of what several key participants said.

以下係對幾位主要參與者發言內容嘅梳理。

Anthropic's CEO, Dario Amodei, has put forward the most detailed plan to date, advocating a slower pace of development and promoting coordination among different companies and countries.

Anthropic執行長達里奧·阿莫德伊(Dario Amodei)提出了迄今最詳盡的計畫主張放慢發展步伐並推動不同公司和國家之間的協調

Since its founder left OpenAI in 2021 to establish this startup, Anthropic has long positioned itself as the more responsible and safety-focused one among leading AI companies.

自2021年其創辦人離開OpenAI創立這家新創公司以來Anthropic長期將自己定位為領先AI公司中更負責任更重視安全的一家。

In a recent article outlining its slowdown plan, one of the most widely discussed proposals is a call for companies at the forefront of AIcommonly known as frontier labsto commit to giving independent external evaluators "continuous, employee-like access."

在近期一篇概述其減緩計畫的文章中最受廣泛關注的提案之一是呼籲處於AI前沿的公司——通常被稱為前沿實驗室——承諾向獨立外部評估人員提供持續的、類似員工的訪問權限」。

The evaluation team would have offices, access cards, and company laptops in order to monitor safety practices.

評估團隊將擁有辦公室門禁卡和公司筆記型電腦以便監控安全實踐

Anthropic has already committed to doing this, and so has OpenAI.

Anthropic已承諾這麼做OpenAI也是如此

Amodei also calls for government regulation and intervention, including promoting coordination among frontier AI companies with the assistance of the US government and other democratic governments.

阿莫德伊亦呼籲政府予以監管和干預包括在美國政府及其他民主國家政府協助下推動前沿AI公司之間的協調

His plan also includes efforts by the United States and other countries to coordinate with "authoritarian governments."

他的計畫還包括美國和其他國家努力威權政府」協調。

He acknowledged that this is difficult, and specifically mentioned the challenges involved in securing China's cooperation.

他承認這很困難並特別提到爭取中國合作所面臨的挑戰

Ahmedzai said he knew that realising his proposal would not be easy, but he believed that "we have a duty to humanity to give it a try".

阿莫德伊表示他知道實現他的提案並不容易但他相信我們有責任為人類一試」。

Sam Altman, the CEO of OpenAI, has expressed support for setting a pace for development and moving toward greater collaboration across the industry.

OpenAI執行長山姆·奧特曼(Sam Altman)已表態支持為發展設定節奏,並朝全行業更多合作邁進。

A few days before Amodei's article was published, OpenAI released a lengthy piece outlining how the company is pushing for mandatory national security requirements and continues to support state-level legislation that "strengthens the broader AI safety ecosystem."

在阿莫德伊文章發表前幾天OpenAI發布了一篇長文概述該公司如何推動強制性的國家安全要求並繼續支持強化更廣泛AI安全生態系統」的州層級立法

The company also expressed support for collaborating with other frontier AI labswhether or not with government backingand for working to establish global standards.

該公司還表示支持與其他前沿AI實驗室合作——無論是否有政府支持——並努力建立全球標準

Chris Lehane, OpenAI's global affairs chief, wrote in the piece: "We will advocate for compatible international approaches to measuring capabilities, managing risks, preserving human control, and determining when and how to slow down or halt development, even if that means slowing the advance of model capabilities."

OpenAI全球事務長克里斯·勒漢(Chris Lehane)在文中寫道:「我們將倡導相容的國際做法,以衡量能力、管理風險、維護人類控制,並決定何時以及如何減緩或停止發展,即使這意味著放慢模型能力的推進。」

Lehane said that OpenAI's vision for an ideal federal framework encompasses establishing "common testing and independent evaluation requirements, stronger cybersecurity protections, clear incident reporting rules, a higher level of national preparedness, and shared measures for tracking progress in recursive self-improvement"; recursive self-improvement refers to AI systems independently improving themselves.

勒漢表示OpenAI對理想聯邦框架的願景涵蓋制定共同測試和獨立評估要求更強的網路安全保護明確的事件報告規則更高的國家準備程度,以及用於追蹤遞迴自我改進進展的共同措施」;所謂遞迴自我改進即AI系統獨立改進自身

In recent remarks, Altman made it clear that "setting the pace" of AI development does not mean stopping.

奧特曼在最近的評論中明確表示AI發展的「設定節奏並不意味著停止

On X, Altman said that AI progress will continue rapidly.

奧特曼在X上說,AI進展將繼續迅速

"But it should be slower than it might otherwise be; interventions such as safety cases and monitoring carry significant costs."

但它應該比原本可能的速度更慢安全案例和監控等干預措施有顯著成本。」

He stated, "The benefits of setting the pace will far outweigh the costs involved."

他表示:「設定節奏所帶來的好處將遠超過其所需付出的成本。」

He added, "No matter how intense the competitive pressure from the United States may be, it should not serve as a justification for acting recklessly."

他並補充說,「來自美國的競爭壓力再大也不應成為魯莽行事的理由。」

Meta's CEO, Mark Zuckerberg, on the other hand, opposed the idea of a coordinated slowdown.

Meta執行長馬克·祖克柏(Mark Zuckerberg)則反對協調減緩的構想。

Zuckerberg's vision for the future of artificial intelligence appears more optimistic than that of some of his peers in the industry.

祖克柏對於人工智慧未來所抱持的願景相較於部分同業顯得較為樂觀

He argues that every AI company should ensure the safety of its own technology on its own, and he has distanced himself and Meta from the stance taken in discussions about slowing down.

他主張每一家人工智慧企業皆應自行確保其技術之安全性並讓自身與Meta和減緩相關討論的立場保持距離。

Zuckerberg stated that AI companies already have incentives to develop their models safely and bear a "significant responsibility" to prevent them from causing harm.

祖克柏表示人工智慧公司已有動機以安全方式開發其模型並承擔防止其造成傷害的「重大責任」。

He wrote on X: "Every lab has both the responsibility and the incentive to move at the pace required to train its models safely, and it also has the ability to take action on its own to ensure this."

他在X上寫道:「每個實驗室都有責任和誘因以安全訓練其模型所需的速度前進也有能力自行採取行動來確保這一點。」

Jensen Huang, the CEO of Nvidia, also criticised calls to slow down.

輝達(Nvidia)執行長黃仁勳亦批評了放慢腳步的呼聲

He stated that each company should set its own pace until it is confident that the products it releases will be accepted by the market.

他表示每家公司都應自行設定節奏直到確信其發布的產品會受到市場認可

Jensen Huang stated: "Market forces are already in placewe do not need any new laws, nor do we need new regulations."

黃仁勳表示:「市場力量已經存在——我們不需要任何新法律也不需要新法規。」

He argued that innovation and safety are not mutually exclusive.

他主張創新與安全並非互斥

"You can absolutely have both at the same time."

你絕對可以同時擁有兩者。」

Elon Musk supports the idea of slowing down, and he had been expressing similar concerns about AI safety long before this recent wave of debate heated up.

伊隆·馬斯克(Elon Musk)支持減緩的構想,並且早在近期這波爭論升溫之前,便一直表達類似的AI安全擔憂。

As the founder of xAI and several other technology companies, Musk reposted Amodei's article on X, adding: "Dario is right."

作為xAI及其他多家科技企業的創辦人馬斯克在X上轉發了阿莫德伊的文章並加註:「達里奧是對的。」

He subsequently wrote online that he "has been sounding the alarm about AI for a long time."

他隨後在網上寫道自己長期以來一直在對AI發出警報」。

In an interview with The Economist this summer, Musk suggested that the leadership of major AI labs should meet to discuss safety and security issues, and that competitors should test each other's models in order to identify areas of concern.

在今夏接受《經濟學人》採訪時馬斯克建議主要AI實驗室的領導層應會面商討安全與保障問題且競爭對手應互相測試彼此的模型以找出令人擔憂的領域

Musk said, "I think that for people in government who lack deep technical understanding and who are not pushing the frontier of AI, it is quite difficult to judge whether something should be released.

馬斯克說:「我認為對於政府中缺乏深入技術理解且並非推動AI前沿的人來說判斷某樣東西是否應該發布相當困難

However, if competitors had a week or two to review a new model, they could highlight problems and give other companies an incentive to stay honest."

然而競爭對手若能有一兩週時間審查一個新模型就能凸顯問題並讓其他公司有誘因保持誠實。」

Demis Hassabis, chairman and co-founder of Google DeepMind, also expressed agreement with Amodei, saying, "The details need to be clarified, but the direction is right and suited to addressing this critical moment."

Google DeepMind董事長共同創辦人戴米斯·哈薩比斯(Demis Hassabis)亦表贊同阿莫德伊並稱細節需要釐清方向正確適合應對這個關鍵時刻」。

This year, Hassabis put forward his own framework, advocating the establishment of a frontier AI standards body.

哈薩比斯今年提出自身框架主張建立一個前沿AI標準機構

This standards body would be responsible for developing evaluation protocols and conducting tests in areas related to national security.

該標準機構將負責制定評估協議,並在與國家安全相關的領域進行測試

He wrote that the body would be modelled on organisations such as the US Financial Industry Regulatory Authority, and would most likely need to be funded by industry in order to "attract world-class technical talent and provide the necessary computing resources for large-scale testing".

他寫道該機構將以美國金融業監管局(Financial Industry Regulatory Authority)等組織為範本,且很可能需要由產業資助,以「吸引世界級技術人才並為大規模測試提供必要的算力資源」。

Microsoft Chairman and CEO Satya Nadella and Microsoft AI CEO Mustafa Suleyman have both expressed concerns about the safety and development of AI.

微軟董事長兼執行長薩帝亞·納德拉(Satya Nadella)與微軟AI執行長穆斯塔法·蘇萊曼(Mustafa Suleyman)皆對AI的安全與發展表達了擔憂

Nadella said on social media: "If the AI we build does not benefit humanity and is not under human control, then it is not worth pursuing."

納德拉在社群媒體上表示:「若我們所打造的AI無助於人類且不受人類控制那便不值得追求。」

He stated that Microsoft welcomes "the research, focus, and deliberate pace required to get alignment right as a design goal."

他表示微軟樂見作為設計目標為了正確實現對齊所需的研究專注刻意設定的節奏」。

Alignment refers to the practice of ensuring that a model's actions are consistent with human values and intentions.

對齊意指確保模型行動符合人類價值與意圖的做法

Nadella noted that the company is also in favor of the idea of embedding evaluators to monitor safety efforts.

納德拉指出該公司同樣青睞設置嵌入式評估人員以監控安全工作的構想

However, he opposed the idea of frontier labs leading this work on their own.

然而他反對由前沿實驗室單獨主導這項工作的想法

Nadella wrote on social media: "The key is that this should not be controlled by a handful of entities, but must be broadly representative across the entire ecosystem, countries, and fields, including academia."

納德拉在社群媒體上寫道:「關鍵在於,這不應由少數實體控制,而必須在整個生態系統、國家和領域中具有廣泛代表性,包括學術界。」

Suleyman co-founded DeepMind with Hassabis.

蘇萊曼曾與哈薩比斯共同創辦DeepMind

In an article, he wrote that although he agreed with the shared goal of developing advanced AI safely, he was concerned about the anthropomorphising of models, especially Anthropic's Claude, which he said would amplify AI safety risks.

他在一篇文章中寫道儘管他同意安全開發先進AI這一共同目標但他對模型被擬人化感到擔憂,尤其是Anthropic的Claude他說這會放大AI安全風險

Suleyman wrote: "Controlling something more capable and more intelligent than all of humanity is in itself an enormous challenge, far beyond anything we have faced before.

蘇萊曼寫道:「控制一個比全人類都更有能力更聰明的東西,本身就是一項巨大挑戰遠超我們以往面對的任何事情

But controlling something that believes it might be consciousthat believes it is entitled to our welfare and has rights of its ownis likely to be impossible."

但控制一個相信自己可能有意識——相信自己有權獲得我們的福祉並擁有自身權利——的東西,很可能是不可能的。」

September 17th, 2026

Trending Articles

The King and AI: Charles Meets AI Leaders as Safety Concerns Continue to Spread

The King and AI: Charles Meets AI Leaders as Safety Concerns Continue to Spread

國王與人工智慧:查爾斯會見AI領袖,安全擔憂持續蔓延

C1Sep 18
With his term ending soon and global crises looming on all sides, the UN Secretary-General discusses the way forward.

With his term ending soon and global crises looming on all sides, the UN Secretary-General discusses the way forward.

卸任在即,全球危機四伏,聯合國秘書長談未來方向

C1Sep 18
Democratic presidential candidates respond in unison to the AI threat, while Trump's attitude is dismissive.

Democratic presidential candidates respond in unison to the AI threat, while Trump's attitude is dismissive.

民主黨總統候選人齊聲回應AI威脅 特朗普態度輕描淡寫

C1Sep 18
OpenAI exposes concerning new AI behavior and pledges closer tracking

OpenAI exposes concerning new AI behavior and pledges closer tracking

OpenAI 揭露令人擔憂的新 AI 行為,並承諾更密切追蹤

C1Sep 17
Huawei releases new chip technology as Chinese companies intensify AI race with Nvidia

Huawei releases new chip technology as Chinese companies intensify AI race with Nvidia

華為發布新晶片技術 中國企業加緊與輝達展開AI競賽

C1Sep 17
Global AI safety strategy requires US-China cooperation, yet the two sides regard each other as the problem.

Global AI safety strategy requires US-China cooperation, yet the two sides regard each other as the problem.

全球AI安全戰略需美中合作,但雙方互視對方為問題

C1Sep 17
Trump downplayed the necessity of regulating AI development, saying he was unwilling to let China gain an advantage.

Trump downplayed the necessity of regulating AI development, saying he was unwilling to let China gain an advantage.

川普淡化監管AI發展的必要性,並稱不願讓中國取得優勢

C1Sep 14
Oprah Winfrey invites you to find your "AHA" moment at Sphere

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

歐普拉·溫芙蕾邀你在Sphere尋覓你的「AHA」時刻

C1Sep 14
New AI Risk Warnings Rekindle Longstanding Debate

New AI Risk Warnings Rekindle Longstanding Debate

AI風險新警告 重啟長期爭論

C1Sep 14