Scott Bessent confirms weekend meeting with He Lifeng days before Xi’s US visit

US Treasury Secretary Scott Bessent on Tuesday confirmed that he will meet Chinese Vice Premier He Lifeng this weekend for the final preparatory talks between the two sides, less than a week before Chinese President Xi Jinping travels to Washington for a bilateral summit with US President Donald Trump .

“With China, we have had some very good private discussions, and I look forward to those continuing this weekend when I meet my Chinese counterpart Vice [Premier] He Lifeng,” Bessent said during a House Financial Services Committee hearing.

Bessent was responding to a question about the alleged role of Chinese banks in facilitating commercial transactions with Iran and whether Washington planned to take action against them.

Bessent is expected to discuss the likely economic outcomes of the Trump-Xi summit and begin preparatory dialogue on artificial intelligence safety and governance.

Last week, the South China Morning Post reported that Washington and Beijing are set to fold the long-awaited AI dialogue into a broader economic meeting.

The two leaders agreed in Beijing in May to establish an intergovernmental dialogue on AI , which ran separately from the economic talks led by Bessent and He for months. The AI channel has struggled to make progress, with basic details of the planned talks still unresolved as Xi’s visit approaches.

Bessent set to address AI and economic issues in meeting with He

Sources said Bessent is now expected to address both AI and economic issues in a single meeting with He, rather than hold separate high-level talks. Bessent leads the US side in both channels, while the tight schedule ahead of the summit has made arranging a second meeting increasingly difficult.

The latest arrangements mirror those in May, when Bessent and He held talks in Seoul just days before Trump met Xi in Beijing .

Another source familiar with the discussions said the upcoming AI talks are expected to be “fairly preliminary” and more about “agenda setting” than grappling with complex issues.

On Tuesday, Bessent also weighed in on the AI safety debate, highlighting the tension between addressing concerns about the risks posed by increasingly powerful AI systems and maintaining US competitiveness with China in the race for AI leadership.

“It does matter whether the good guys or the bad guys have this,” Bessent said in response to Massachusetts Democrat Stephen Lynch, who argued it probably wouldn’t matter whether the good or bad guys developed superintelligence if it couldn’t be controlled.

“As a person who manages the relationship with China on AI, it matters a great deal,” Bessent added.

Trump on Monday dismissed growing concerns about AI, claiming America already has guardrails to regulate the industry and warning that casting doubt on development only benefits China .

“The only control or ‘guardrails’ ‌that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” Trump wrote on social media.

“We already have tremendous CRIMINAL and REGULATORY power over these ⁠companies! There is a SICK conspiracy going on against AI and Data ‌Centers, and the only one that is happy about it is China.”

Bessent echoed Trump’s stance on Tuesday, arguing that “AI is the most important technology of this century, and that we must maintain US leadership”.

Bessent’s testimony came on the same day that Senator Bernie Sanders and former Trump White House aide Steve Bannon joined forces at a Washington conference, dubbed the Pro-Human Assembly , to call for new policies to curb the rapidly accelerating development of AI.

The AI safety issue gained prominence last week when a former Anthropic employee warned that the technology may lead to human extinction in the next decade.

On Saturday, Anthropic CEO Dario Amodei proposed a three-step framework to slow the pace of AI development and give policymakers and companies more time to address the risks posed by rapidly advancing technology.

The proposal quickly won support from other prominent industry leaders, including xAI chief Elon Musk and OpenAI CEO Sam Altman .

However, the Trump administration is prioritising AI competition with China as a key driver of AI development and is unwilling to slow down.

Concerns rise over Chinese AI models using ‘distillation’ from US rivals

The Treasury Secretary once again raised concerns about Chinese AI models using “distillation” from US models, which he described as one area where the US could challenge Beijing.

“One of the ways we can push back against China, because the Chinese models distil from the US models, which is a polite scientific word for steal,” he said.

Model distillation is a widely used AI training technique in which a smaller model learns from the output of larger ones.

Last week, three US agencies – the National Security Agency, the Cybersecurity and Infrastructure Security Agency and the FBI – accused leading Chinese AI companies of engaging in “aggressive, malicious and targeted distillation activities at an industrial scale”. They alleged that the practice was carried out “likely with Chinese government awareness”.

During an afternoon press conference at the White House on Tuesday, US Attorney General Todd Blanche was asked whether the US Justice Department would “go after” China over the alleged distillation or whether Trump would bring it up when he meets with Xi next week.

“When you say China, you’re talking about an entire country, and so those two presidents have a great relationship, and they get along, they make deals together,” said Blanche.

“Are there separate issues with actors in China and companies in China and bad actors? That has existed long before this administration and will exist for the next several years.”

Blanche said the US Justice Department and other executive branch agencies would continue to combat theft “in the right way”.

“I don’t think it’s fair to just label China as a country,” he added. “There’s bad actors in China that are doing it, which we’re definitely going to go after.”

The Chinese embassy in Washington rejected US allegations, terming distillation “a common technical practice used by AI developers worldwide, including US companies”.

“The US allegations are unfounded and politicise a normal technological and commercial practice,” embassy spokesman Liu Chang told the SCMP.

As Chinese companies have narrowed the capability gap in frontier AI models, US firms and Trump administration officials have increasingly accused Beijing of stealing and copying American technology to speed up the development of its own systems. -- SOUTH CHINA MORNING POST

Thank you for your report!

Trump says China is driving anger at US data centres. These residents disagree, reports SCMP

SMG, SCMP ink MOU aimed at strengthening business ties between M'sia, Hong Kong

SMG and SCMP ink landmark deal

美国财政部长斯科特·贝森特周二证实，他将于本周末与中国国务院副总理何立峰会面，进行双方的最后准备会谈。不到一周后，中国国家主席习近平将前往华盛顿与美国总统唐纳德·特朗普举行双边峰会。

贝森特在众议院金融服务委员会听证会上表示：“我们与中国进行了一些非常好的私下讨论，我期待本周末与中国副总理何立峰会面时继续进行这些讨论。”

贝森特是在回答有关中国银行在促成与伊朗的商业交易中扮演的角色，以及华盛顿是否计划对它们采取行动的问题时做出上述表态的。

预计贝森特将讨论特朗普-习近平峰会可能带来的经济成果，并开始就人工智能安全和治理问题展开准备性对话。

上周，《南华早报》报道称，华盛顿和北京计划将期待已久的人工智能对话纳入更广泛的经济会议中。

两位领导人5月在北京同意建立人工智能政府间对话机制，该机制独立于贝森特和贺建奎主持的经济会谈之外，独立运作了数月。人工智能对话渠道进展缓慢，随着习近平访华临近，计划会谈的基本细节仍未敲定。

贝森特将在与贺建奎的会晤中讨论人工智能和经济问题。

消息人士称，贝森特预计将在与贺建奎的一次会晤中同时讨论人工智能和经济问题，而不是分别举行高级别会谈。贝森特在这两个渠道都代表美方，而峰会前夕的紧凑日程使得安排第二次会晤变得越来越困难。

最新的安排与 5 月份的安排类似，当时贝森特和贺建奎在首尔举行了会谈，几天后特朗普在北京会见了习近平。

另一位熟悉讨论情况的消息人士称，即将举行的人工智能会谈预计将是“相当初步的”，更多的是关于“议程设置”，而不是探讨复杂的问题。

周二，贝森特也对人工智能安全辩论发表了看法，强调了在解决日益强大的人工智能系统带来的风险担忧与保持美国在人工智能领导地位竞赛中与中国竞争之间的紧张关系。

“好人还是坏人拥有这种能力确实很重要，”贝森特回应马萨诸塞州民主党人斯蒂芬·林奇时说道。林奇认为，如果超级智能无法控制，那么好人还是坏人发展出超级智能可能都无关紧要。

“作为负责与中国在人工智能领域关系的人，这非常重要，”贝森特补充道。

周一，特朗​​普驳斥了人们对人工智能日益增长的担忧，声称美国已经有了监管该行业的保障措施，并警告说，对人工智能发展表示怀疑只会让中国受益。

特朗普在社交媒体上写道：“人工智能唯一需要的控制或‘护栏’就是一位强大而睿智（高智商！）的总统，而美国在这方面人才济济！”

“我们已经对这些公司拥有巨大的刑事和监管权力！一场针对人工智能和数据中心的邪恶阴谋正在进行，而唯一对此感到高兴的只有中国。”

贝森特周二呼应了特朗普的立场，认为“人工智能是本世纪最重要的技术，我们必须保持美国的领导地位”。

贝森特作证的同一天，参议员伯尼·桑德斯和前特朗普白宫助理史蒂夫·班农在华盛顿联合召开了一场名为“亲人类大会”的会议，呼吁制定新政策来遏制人工智能的快速发展。

上周，一位前人智库员工警告称，人工智能安全问题可能在未来十年内导致人类灭绝，这一警告使人工智能安全问题备受关注。

周六，Anthropic 首席执行官 Dario Amodei 提出了一个三步框架，旨在减缓人工智能的发展速度，让政策制定者和企业有更多时间来应对快速发展的技术带来的风险。

该提议很快赢得了其他知名行业领袖的支持，包括 xAI 首席执行官埃隆·马斯克和 OpenAI 首席执行官萨姆·奥特曼。

然而，特朗普政府将与中国的AI竞争视为AI发展的关键驱动力，并优先考虑这一竞争，且不愿放慢步伐。

人们越来越担心中国人工智能模型会借鉴美国竞争对手的“技术提炼”方法。

财政部长再次对中国人工智能模型使用美国模型的“提炼”方式表示担忧，他称这是美国可以挑战北京的一个领域。

他说：“我们可以反击中国的一种方式，因为中国的模式是从美国的模式中提炼出来的，而‘提炼’用委婉的科学术语来说就是‘窃取’。”

模型蒸馏是一种广泛使用的人工智能训练技术，其中较小的模型从较大模型的输出中学习。

上周，美国国家安全局、网络安全和基础设施安全局以及联邦调查局这三个机构指控中国领先的人工智能公司从事“大规模的、具有侵略性的、恶意的、有针对性的技术提炼活动”。他们声称，这种做法“很可能是在中国政府的默许下进行的”。

周二下午在白宫举行的新闻发布会上，美国司法部长托德·布兰奇被问及美国司法部是否会就涉嫌蒸馏事件“追究”中国的责任，或者特朗普是否会在下周与习近平会面时提出此事。

“当你谈到中国时，你指的是整个国家，所以这两个国家元首关系很好，他们相处融洽，一起达成协议，”布兰奇说。

“中国境内的演员、公司和不良行为者之间是否存在不同的问题？这个问题在本届政府执政之前就已存在，而且在未来几年内仍将存在。”

布兰奇表示，美国司法部和其他行政部门将继续“以正确的方式”打击盗窃行为。

他补充说：“我认为仅仅将中国视为一个国家是不公平的。中国境内有一些不法分子在这样做，我们一定会追究他们的责任。”

中国驻华盛顿大使馆驳斥了美国的指控，称提炼是“包括美国公司在内的世界各地人工智能开发人员使用的常见技术做法”。

“美方的指控毫无根据，并将正常的科技和商业行为政治化了，”大使馆发言人刘畅告诉《南华早报》。

随着中国企业在尖端人工智能模型方面的能力差距不断缩小，美国企业和特朗普政府官员日益指责北京窃取和抄袭美国技术，以加速自身系统的研发。——《南华早报》

感谢您的报告！

特朗普称中国煽动民众对美国数据中心的愤怒。据《南华早报》报道，当地居民对此表示反对。

SMG与SCMP签署谅解备忘录，旨在加强马来西亚与香港之间的商业联系

SMG和SCMP签署里程碑式协议