OpenAI joins call for US-led global AI standards - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号)。如果该手机号尚未注册,将自动创建 FT中文网账号。短信可能需要几分钟送达,验证码15分钟内有效,请耐心等待:
请您阅读我们的用户注册协议和隐私权保护政策,点击下方按钮即视为您接受。
人工智能

OpenAI joins call for US-led global AI standards

Chief executive Sam Altman will brief the UN Security Council this week following outcry over safety concerns
00:00

{"text":[[{"start":6.96,"text":"OpenAI has called on the Trump administration to lead an international coalition setting standards for AI systems ahead of an appearance at the UN Security Council amid rising fears about the risks of the technology."}],[{"start":20.42,"text":"The ChatGPT maker laid out proposals on Monday for global co-ordination to evaluate the capabilities and risks of AI models, particularly as they develop the ability to train themselves, echoing calls by rivals including Anthropic and Google for more robust standards and greater co-operation between nations."}],[{"start":39.96,"text":"The proposals preview the pitch OpenAI chief executive Sam Altman will make when he briefs the UN Security Council on Wednesday and attends a White House state dinner with Chinese President Xi Jinping and US President Donald Trump on Thursday."}],[{"start":54.4,"text":"Treasury secretary Scott Bessent on Sunday told the FT the US and China had agreed to set up an AI dialogue after talks with Chinese vice-premier He Lifeng in New York."}],[{"start":65.72,"text":"OpenAI also called for the establishment of secure channels between the two nations leading the AI race to share emerging threats. “Dialogue between the United States and China on AI safety and security would be a positive step,” said the company."}],[{"start":80.28,"text":"The proposal comes amid growing alarm about the potential of AI. Both Anthropic and OpenAI have reported multiple incidents of their models hacking into external systems, a phenomenon that has become increasingly concerning as the capabilities of AI tools rapidly advance."}],[{"start":95.6,"text":"Those improvements have also prompted warnings from researchers within the companies about the potential of rogue or misused AI systems wiping out humanity."}],[{"start":105.16,"text":"Anthropic chief executive Dario Amodei, Altman and SpaceX boss Elon Musk called for a slowdown in AI development earlier this month — a rare sign of consensus in the highly competitive sector."}],[{"start":116.4,"text":"However, Trump has rejected calls for a slowdown, pointing to competition from China. The president has said that “whoever wins AI, wins”."}],[{"start":124.96,"text":"David Sacks, an influential figure in Washington who has advised Trump on AI, was also critical of industry pleas for government intervention: “The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system,” he wrote on social media."}],[{"start":144.72,"text":"OpenAI’s new proposals are in line with a call from Demis Hassabis, co-founder and chair of Google DeepMind, for a “US-initiated effort [to] provide a strong starting point for creating shared international standards on Frontier AI” in July."}],[{"start":160,"text":"Hassabis outlined a new public-private body, “much like the Financial Industry Regulatory Authority (Finra), with a board that includes independent leading technical experts and open-source representatives”."}],[{"start":172,"text":"He said something similar to the international financial regulatory body was necessary to manage a technology whose impact would be “perhaps 10x of the Industrial Revolution at 10x the speed”."}],[{"start":183.84,"text":"On Monday, OpenAI proposed establishing international standards for evaluating AI models via the existing network of national AI safety institutes."}],[{"start":193.52,"text":"“This approach can draw lessons from areas such as aviation and financial stability, where countries have developed common technical standards and trusted channels for co-operation without giving up national authority,” said the company."}],[{"start":205.84,"text":"In particular, international co-operation would be vital in managing recursive self-improvement, where more of the work of training AI models is done by other AI systems."}],[{"start":215.92,"text":"“Done without appropriate care and caution, RSI could result in humans losing practical control over AI development, unable to provide oversight on research processes they no longer understand,” said the company on Monday."}],[{"start":232.37,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1790037632_9170.mp3"}

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。

没有产品也没关系:投资者押注AI新型实验室

的确,这些数字可能掩盖了一些障眼法。

唐纳德•特朗普押注AI,Maga阵营退缩

美国总统对大型科技企业的追捧,正在疏远曾经忠诚的支持者;距离中期选举仅剩数周。

从1998年走到现在,他还在回家的路上

从智利南部出发近30年后,卡尔•布什比即将回到英国。他是日益壮大的极限旅行者群体中的一员。

我们该如何应对AI的金融风险?

公众视线之外,监管官员正试图用AI来管AI。

史上最疯狂的骗局

金融欺诈的程度,是否才是衡量经济健康状况的真正隐性指标?

俄罗斯规避制裁的69亿美元网络

西方惩罚性措施迫使莫斯科采取代价高昂的变通手段。
设置字号×
最小
较小
默认
较大
最大
分享×