反贼网
Machine translation · the Chinese original is authoritative · View original

Voice of America Chinese Radio Broadcast (September 15, 2026)

1# · OP Author:美国之音中文网 Published:2026-09-14 22:00 Replies:0 Views:8 Permalink:fanzei.net/d_ufavsr

This episode of Voice of America Chinese broadcast covers Trump's statements on artificial intelligence and Russia-Ukraine energy targets, developments in US-Iran relations, the New Delhi Declaration of the 18th BRICS summit, Pentagon officials attending the Beijing Xiangshan Forum, Iran's capture of a US-made unmanned underwater vehicle, an Anthropic report accusing Chinese companies and the government of model data extraction and overseas monitoring, an academic network's call for the release of US-Myanmar scholar Min Htin, and the head of Britain's MI5 warning of strategic security threats posed by China.

Video link:

Full text:

Hello everyone, listeners and friends. Welcome to Voice of America's Chinese broadcast. I am Xu Xiangyun reporting to you from Washington, D.C., the capital of the United States. The main stories on today's program include: President Trump reaffirming the necessity for the United States to be the world leader in artificial intelligence while downplaying concerns over the rapid development of the field; Donald Trump stating that the Pentagon has confirmed that a senior official from the US Department of War responsible for China, Taiwan, and Mongolia affairs will attend the Xiangshan Forum opening in Beijing on September 15; and the 18th BRICS summit releasing the New Delhi Declaration, calling for maximum restraint in the Middle East and avoiding actions that could worsen the situation.

US President Donald Trump reiterated the necessity for the United States to be the world leader in artificial intelligence, downplaying concerns over the rapid development of the field despite calls from AI industry leaders to slow down progress to ensure safety and security. Trump told reporters on Sunday that the United States is ahead of China in artificial intelligence and that he wants the United States to be the world leader, because whoever wins artificial intelligence wins it all. Trump said we can set up safety guardrails, we can do this and that, but I think there are a lot of negative voices that shouldn't be raising this question constantly bringing it up, and the things they are proposing simply won't happen. He added on Truth Social on Monday that we already have tremendous formal and regulatory power over these companies, and some people are running a morbid conspiracy about AI and data centers, with the only ones happy about this being China. Artificial intelligence will likely be one of the topics discussed when Trump meets with Chinese leader Xi Jinping in Washington later this month.

Chinese Foreign Ministry spokesperson Guo Jiakun stated at a Monday press conference that the development of artificial intelligence should be carried out in an open and inclusive manner. He said that spreading various threat narratives, engaging in confrontation and vicious competition will only disrupt the global AI governance process and is not in the interest of any party. The comments from US and Chinese officials followed an open letter published on Saturday by Dario Amodei, head of US AI giant Anthropic. Amodei said that if China takes the lead in AI development, it would pose a grave danger to the United States and the entire world, and he proposed a series of measures that he believes would both ensure US leadership in AI and help mitigate potential security risks. He called for banning the sale of high-performance AI chips and other equipment to China and strengthening the security of AI companies to prevent potential intellectual property theft. Amodei also said US AI companies should allow independent evaluators to access their systems to monitor potential security risks and coordinate with other democratic countries to promote safe AI development while maintaining an advantage over China's AI industry. Sam Altman, head of another AI company, OpenAI, welcomed Amodei's proposals, saying that slowing down development does not mean stopping AI progress, and avoiding costly negative outcomes is well worth it.

President Donald Trump said that both Ukraine and Russia have agreed to stop striking each other's energy targets. Trump wrote on Truth Social on September 14 that Ukraine has agreed not to strike Russian energy targets and Russia has agreed to do the same. The global rise in diesel prices is mainly caused by the Russia-Ukraine war, not Iran. Previously, Trump called on Ukraine on Sunday to halt attacks on Russian refineries, telling reporters that Ukrainian President Volodymyr Zelenskyy must stop damaging Russia's diesel supply, adding that there are many other targets and these attacks are hurting the world. Ukraine stated that Russian refineries are legitimate military targets because the oil industry funds Moscow's war. Ukraine and Russia have not yet commented on Trump's Truth Social post on Monday. In recent weeks, both sides have stepped up attacks on each other's energy and economic infrastructure, and Russia has also begun attacking border crossings between Ukraine and its Western allies. NATO Secretary General Mark Rutte criticized the drone attacks launched by Russia near the Ukraine-Poland border and stated that this would prompt NATO to increase its support for Kyiv. Speaking at NATO headquarters in Brussels on September 14, Rutte said these attacks near NATO territory show the desperation of Russian President Vladimir Putin and his desire to spread fear and terror.

US President Donald Trump stated on Monday that Iranian leaders want to reach a deal with the United States and that the US is open to it. President Trump posted on Truth Social that a struggling Iran is eager to make a deal, and as for whether the US chooses to intervene, I will decide, though we are open to the idea. In another Truth Social post, President Trump noted that energy supplies are still moving through the Strait of Hormuz, while stating that countries dependent on Gulf oil should pay the United States to compensate for the efforts made by the US military to guide and assist merchant ships passing through the waterway. President Trump said oil is flowing through the Strait of Hormuz, and after this sham conflict ends, countries around the world that have never provided us with any help should and will compensate the US financially; we do this more for others than for ourselves, and this has been going on for generations. The United States and Iran have accused each other of violating the interim agreement signed on June 18 under the mediation of Pakistan and Qatar. President Trump announced the end of the ceasefire with Iran on July 7 after the Iranian military attacked three ships passing through the Strait of Hormuz, and negotiations have stalled since then. Just days after the attacks, the US resumed a naval blockade of Iranian ports. Meanwhile, Tehran rejected negotiations over its nuclear program, turning instead to talks with Oman regarding the management of the Strait of Hormuz. Qatar, one of the mediators, called on the US and Iran last month to return to the previously signed memorandum of understanding and seek an agreement on the strait, with Doha stating this would allow us to return to the negotiating framework.

The two-day BRICS summit in New Delhi concluded on September 13, with host nation India demonstrating its coordination and diplomatic skills, not only producing a summit declaration acceptable to the warring parties in the Middle East, but also maintaining the BRICS positioning as non-Western rather than anti-Western. The 18th BRICS summit issued the New Delhi Declaration, calling for maximum restraint in the Middle East and avoiding actions that could worsen the situation. The Middle East war was undoubtedly the most watched issue at this BRICS leaders' summit, and internal divisions within the expanded BRICS group followed. The smooth release of the New Delhi Declaration was seen by analysts as a major diplomatic achievement for India. Iran, which became a BRICS member in 2024, along with the United Arab Emirates and Saudi Arabia, stand on opposite sides in this Middle East war. Zheng Jiapin, associate professor at Shiv Nadar University in India, told VOA that given the grim global situation, India's ability to bring these countries together on one platform is an achievement in itself. Fang Tianci, director of the Center for Indian Studies at National Tsing Hua University in Taiwan, told VOA that India has essentially accomplished this task. The BRICS countries include Iran, China, and Russia, which carry strong anti-US sentiments, as well as the UAE and Saudi Arabia, which are relatively close to the United States. Lin Xiaozhen, assistant professor at the Graduate Institute of International Affairs and Strategy at Tamkang University, stated in written comments to VOA that although China holds the largest

its structural influence still cannot unilaterally determine the group's direction. The New Delhi Declaration expressed concern over the tense situation in the Middle East with relatively neutral wording while condemning attacks on civilian infrastructure and nuclear facilities. Although the declaration is seen as an indirect criticism of the US and Israel, leaving room for all parties without spelling things out serves as a signal of goodwill toward US President Donald Trump.

The Pentagon confirmed to Voice of America that Michael Chase, the Deputy Assistant Secretary of Defense for China, Taiwan, and Mongolia, will represent the Department of Defense at the Xiangshan Forum opening in Beijing on Tuesday. A defense official told VOA via email on Monday that we can confirm that Michael Chase, Deputy Assistant Secretary of Defense for China, Taiwan, and Mongolia, will represent the Department of Defense at this year's Xiangshan Forum. The Pentagon did not specify whether Chase plans to hold bilateral talks with Chinese Defense Minister Dong Jun or other officials of the People's Liberation Army. Reuters previously reported, citing a US official, that Chase is expected to lead the US delegation to the forum, and his attendance level is higher than last year when the US sent a defense attaché from the US Embassy in China to participate. The 13th Beijing Xiangshan Forum is scheduled to be held from September 15 to 17 under the theme of building consensus for stability and jointly promoting security governance. The Chinese Ministry of Defense stated that this forum will discuss issues such as international order and strategic stability, Asia-Pacific security risks, maritime lane security, military applications of emerging technologies, and space security governance. Spokesperson Jiang Bing stated on September 14 that the forum features official delegations, experts, and scholars from nearly 100 countries, regions, and international organizations. Dong Jun is expected to deliver a speech at the opening ceremony. This meeting is held shortly after US Indo-Pacific Command Commander Admiral Samuel Paparo is scheduled to hold his first face-to-face meeting with General Yang Zhibin, commander of the PLA Eastern Theater Command responsible for operations in the Taiwan Strait direction.

The Islamic Revolutionary Guard Corps announced last week that it had captured an American unmanned underwater vehicle in the Strait of Hormuz. US media reported that this vehicle matches the unmanned underwater vehicle produced by the US company Anduril. The US military uses this vehicle in the Strait of Hormuz to perform unmanned maritime surveillance, seabed mapping, and anti-mine warfare, and Taiwan also uses it to inspect submarine cables and protect underwater security. There are outside concerns that Iran will obtain data from this vehicle, but experts said the technology Iran can obtain from this captured malfunctioning vehicle and share with China and Russia is limited. The US military uses Anduril's Dive-LD unmanned underwater vehicle to perform unmanned maritime surveillance, seabed mapping, and anti-mine warfare in the Strait of Hormuz, and this vehicle is also used in waters surrounding Taiwan. According to Taiwan media reports, Taiwan's National Chung-Shan Institute of Science and Technology received Anduril's Dive-LD unmanned underwater vehicle system in August this year and first applied it to submarine cable inspection capabilities to prevent deliberate sabotage. The Islamic Revolutionary Guard Corps announced on September 8 that it had captured an American unmanned underwater vehicle at the entrance of the Strait of Hormuz and released images of the vehicle, which, upon comparison, matches Anduril's Dive-LD unmanned underwater vehicle. US Central Command stated that this unmanned underwater vehicle had malfunctioned more than a day before it was discovered by Iran, and as an older model, it neither collected sensitive data nor carried any classified sonar or radar equipment. Professor Kevin Rowlands, editor at the Royal United Services Institute in Britain, said in an interview with Reuters that unmanned underwater vehicle manufacturers and the US side should have already established relevant protocols to ensure that once the equipment falls into enemy hands, the most critical technology cannot be used. For Iran at the present stage, the hardware might be more valuable, and the engineering technology used in these systems exceeds Iran's current capabilities. While Iran indeed possesses a large number of drones, has provided drones to Russia for the Russia-Ukraine conflict, and is adept at using unmanned systems, in terms of propulsion, guidance, maneuverability, depth control, and specific communication technology, Iran's technical level may not be as advanced as the US system. Therefore, through reverse engineering, they might gain a deep understanding of the US technical path and begin to imitate it. If Iran chooses to share these technologies with Russia or China—and we know such relationships do exist—neither the US nor the rest of us will likely be able to prevent this technology spillage to regions outside Iran. According to Anduril, the Dive-LD is a large-diameter autonomous underwater vehicle measuring 5.8 meters long and 1.2 meters in diameter, capable of autonomously executing missions lasting up to 10 days with a maximum diving depth of up to 6,000 meters. Its shell utilizes 3D printing technology to achieve a high degree of customization, and it adopts innovative manufacturing processes for internal core structural components and buoyancy systems to allow custom modifications for adapting to different types of payloads and sensors, applicable for defense or commercial operations. Anduril also provides enterprise leasing options. Whether in the short or long term, I do not believe this will cause a major impact; in any conflict, we should expect system losses, destruction, or capture, and we simply need to accept this as part of conventional warfare and handle it calmly. According to reviews by Defense News and Taiwan media reports, Taiwan's National Chung-Shan Institute of Science and Technology cooperated with Anduril to carry out adaptive and customized modifications on the Dive-LD according to Taiwan's needs, integrating multi-frequency scanning capabilities to detect mines and submarines, making it a highly customized unmanned underwater vehicle.

On Monday, the three major stock indices saw the Dow Jones Industrial Average fall by 0.29%, the S&P 500 fall by 0.48%, and the tech-heavy Nasdaq Composite fall by 0.56%. From Monday to Friday US Eastern Time at 6:00 PM, and Tuesday to Saturday Beijing Time at 6:00 AM, please tune in to the Voice of America Chinese broadcast and feel free to browse the Voice of America Chinese website voachinese.com and follow our social media platforms such as X and YouTube to get more news and information.

US artificial intelligence company Anthropic stated that over the past eight months, it has blocked model capability extraction operations associated with seven Chinese artificial intelligence laboratories, including operations attributed to Alibaba operators initiating over 151 million interactions with Claude within three months. In its latest artificial intelligence threat intelligence report released on September 10, Anthropic stated it found that operators associated with Chinese enterprises such as Alibaba, Moonshot AI, DeepSeek, and Xiaomi attempted to massively acquire answers, reasoning processes, and other capabilities generated by Claude, and used these data to train their own models. Anthropic blocked operations related to seven Chinese artificial intelligence laboratories in total during the 8-month period covered by the report. Speaking about activities related to China, Anthropic's report stated that operators associated with Alibaba carried out the largest illegal distillation operation discovered by the company during this period. According to Anthropic's statistics, from May to July this year, over 3,500 fake accounts related to Alibaba initiated more than 151 million interactions with Claude, approaching nearly 3 million daily at peak times. Anthropic alleged that the operators requested Claude to display detailed problem-solving reasoning processes and used the obtained outputs to train and improve Alibaba's Tongyi Qianwen series models. This figure is much higher than the scale previously announced by Anthropic. In June, a policy director at the company stated in a letter to US lawmakers that Alibaba-related operators conducted approximately 28.8 million unauthorized interactions with Claude from April 22 to June 5. Anthropic also alleged that Moonshot AI and DeepSeek adopted a method that may trigger data privacy concerns by forwarding real-time dialogues between users and their own chat bots to Claude, having Claude generate responses, and then utilizing these outputs to train their company's models. According to Anthropic, some content submitted by users may contain sensitive information, and the relevant users may not know that the questions they input into Chinese artificial intelligence products were sent to an American company's model for processing. Anthropic stated that Moonshot AI-related systems forwarded approximately 23 million user requests to Claude, while DeepSeek-related systems forwarded about 12.1 million requests, with these two activities totaling over 35 million. If this accusation is true, the risks will involve not only the intellectual property rights and commercial competitive advantages of American artificial intelligence models, but also whether user data was handed over to overseas third parties without sufficient notification or consent. Anthropic's report also expanded China-related risks to state-sponsored surveillance activities, stating that the company discovered multiple account groups associated with the Chinese government using Claude to assist in collecting, organizing, and analyzing information on overseas individuals and groups. One operation attributed by Anthropic to China-related actors focused on collecting data on Uighurs located in Syria, paying particular attention to Uighurs who joined the newly established Syrian army or were related to local armed organizations. Another group of accounts assessed by Anthropic as being associated with the Chinese government used Claude to build personal profiles of Asian religious leaders and members of religious groups. Targets listed in the report included multiple senior Catholic bishops in Asia, leaders of the Presbyterian Church in Taiwan, civil society members of Tibetan Buddhism, and personnel related to the Tibetan exiled administrative system. The report also claimed that other related accounts utilized Claude to collect information on overseas dissidents and overseas Chinese communities. However, these accusations stem from Anthropic based on account activity

, the report derived from network infrastructure and usage patterns did not disclose the true identities of the relevant accounts nor specify which Chinese government department they are affiliated with. The focus of the current controversy is whether certain enterprises use false accounts, third-party interfaces, and proxy services to circumvent geographic and access restrictions, acquire proprietary model outputs on a large scale, and violate terms of service to use them for training competing products.

On September 9, Scholars at Risk, an organization that monitors academic freedom and protects scholars globally, published an open letter on its official website addressed to Chinese President Xi Jinping, urging the Chinese government to immediately release Minit Min, a detained American scholar of Burmese descent. Robert Quinn, executive director of Scholars at Risk, wrote in the email that he was writing on behalf of the organization to express concern over the situation of U.S. citizen and political scholar Minit Min. He stated that U.S. Secretary of State Marco Rubio has determined that Minit Min is wrongfully detained in China and her current whereabouts are unknown. Scholars at Risk urged the Chinese authorities to ensure her immediate and unconditional release. The email stated that on June 3, Minit Min arrived at Kunming Changshui International Airport in Yunnan with a valid visa to attend an academic seminar, at which point she was detained. After losing contact with the outside world for more than a week, Chinese authorities publicly accused her of suspected espionage activities and endangering national security, without disclosing reliable evidence to support the charge. Citing public reports, the email noted that since her detention, U.S. diplomatic personnel have obtained only very limited consular access, while her husband Sylvia and teenage daughter Su—both also U.S. citizens—have been unable to speak with her. Minit Min has been kept in solitary confinement at an undisclosed location under residential surveillance at a designated location. The email pointed out that existing indications suggest Minit Min was detained for peacefully exercising her rights to academic freedom and freedom of expression, rights explicitly protected by international human rights instruments including the Universal Declaration of Human Rights and the International Covenant on Civil and Political Rights. The email stated that the manner of Minit Min's arrest, her unknown whereabouts, and her continued detention have raised serious concerns about whether due process is being implemented and regarding her future well-being and safety. The email emphasized that scholars are deeply concerned not only about Minit Min's own situation but also about the signal her detention sends to American scholars of China. The email stated that scholars firmly believe Minit Min and all scholars—including those at her institution, the Myanmar Institute for Strategic and Policy Studies—should be able to conduct research without the threat of intimidation or arbitrary detention. The email concluded by calling on China to immediately and unconditionally release Minit Min and thanking President Xi Jinping for his attention to this important matter. Copies of the email were also sent to Chinese Foreign Minister Wang Yi, Chinese Ambassador to the U.S. Xie Feng, U.S. Ambassador to China David Puderbaugh, U.S. Secretary of State Rubio, and four U.S. officials at the United Nations. This is the second time in two weeks that Scholars at Risk has spoken out publicly for Minit Min. On August 27, the organization released an open letter signed by more than 100 scholars and professors from the U.S. and other countries expressing grave concern over Minit Min being illegally detained by China while traveling for academic activities. Currently, 113 scholars have signed the open letter to express their support. September 9 is Minit Min's birthday. Minit Min's younger brother, Aung Myo Hein, said in an exclusive interview with Voice of America on August 24, "We really hope my brother can gain his freedom before his birthday." On September 8, the Myanmar Institute for Strategic and Policy Studies, founded by Minit Min, published a commentary on its official website titled "A Birthday Without Him," revealing that whenever Minit Min's birthday arrives, she usually reflects on her life journey and posts her thoughts on social media to share her deep inner reflections.

Ken McCallum, Director General of MI5, wrote an article on Friday, the 25th anniversary of the September 11 terrorist attacks, warning that alongside the risk of large-scale terrorist attacks, the UK also faces strategic security threats posed by powerful nation-states. Just recently in July, newly appointed British Prime Minister Andy Burnham held his first telephone call with Chinese President Xi Jinping. Although McCallum's article published in The Independent did not name any specific country, his previous definition of the China threat and the UK's first trial of a China-related espionage case this year have placed the new government's balancing act between economic engagement with China and national security under scrutiny. According to Downing Street, Burnham discussed pragmatic cooperation to promote economic growth during his call with Xi Jinping, while also raising issues of domestic security, Hong Kong, and human rights. In his October 2025 annual threat report, McCallum stated that over the past year, MI5 had increased by 35% the number of investigations into individuals suspected of participating in state-threat activities. The China-related activities he listed included cyberespionage, covert technology transfer, clandestine interference in public life, and the harassment and intimidation of dissidents. At the same time, he argued that the UK should resist threats while also seizing cooperation opportunities that serve the national interest. In June of this year, Five Eyes security agencies jointly warned that Chinese military intelligence agencies are using professional networking and recruitment sites to approach civil servants, military personnel, journalists, and think tank personnel who have access to sensitive information. However, MI5 did not disclose the proportion of China-related cases nor the personnel and budget allocated to work concerning China. Sam Dunning, director of the China-UK Transparency Project, told Voice of America that if the threat is rising as claimed, the question is whether resources are also increasing and whether existing resources are sufficient. Pro-democracy activist and current UK resident Nathan Law, who is wanted with a bounty by Hong Kong police, said the deepest impact of transnational repression lies in the chilling effect created by suspected surveillance, online harassment, and pressure applied to relatives in Hong Kong. He said a foreign government does not need to successfully monitor everyone; as long as enough people believe they might be monitored and intimidated, part of the objective has been achieved. He believes the central government's awareness of the threat has not yet translated into consistent and effective support for victims by local police. Ben Keith, a barrister who has frequently represented Hong Kong and Chinese dissidents in asylum and Interpol Red Notice cases, told Voice of America that based on the cases he has handled, China-related operations simultaneously deploy internet technology, human intelligence, and diplomatic resources. He said they are willing to invest substantial resources to track individuals who do not actually pose a significant threat; in other words, the objective is often not to steal secrets, but to monitor and suppress those who might embarrass Beijing.

Key events timeline · Selected related nodes
2026年6月18日
US and Iran Sign Interim Agreement, Breaks Down Weeks Later
The United States and Iran signed an interim agreement brokered by Pakistan and Qatar. Subsequently, due to Iranian military attacks on three ships in the Strait of Hormuz, Trump announced the end of the ceasefire with Iran on July 7, leading to a halt in negotiations and the resumption of the naval blockade.
Related topics Trump Artificial Intelligence International Image Washington 科技战
Transcript · timestamped
各位听众朋友大家好欢迎收听美国之音的中文广播我是徐湘芸从美国首都华盛顿向您播报今天的节目今天节目的主要内容有特朗普总统重申了美国在人工智能领域成为世界领导者的必要性同时淡化了人们对该领域快速发展的担忧唐纳德特朗普总统说
五角大楼证实美国战争部负责中国
台湾和蒙古事务的首席官员
将出席9月15日在北京开幕的香山论坛
第18次金砖峰会发布新德里宣言
呼吁中东地区保持最大程度的克制
避免采取可能恶化局势的行动
更多新闻内容欢迎锁定收听
美国之音的中文广播
您现在正在收听的是
听的是美国之音中文广播
美国总统唐纳德·特朗普
重申了美国在人工智能领域
成为世界领导者的必要性
他淡化了人们对该领域
快速发展的担忧
尽管人工智能行业领袖
呼吁放慢进展
以确保安全与保障来听报道
特朗普星期日对记者表示
美国在人工智能方面领先中国
他希望美国成为世界领导者
因为谁赢得人工智能
谁就能赢得胜利
特朗普说我们可以设立安全护栏
我们可以做这个做那个
但我认为有很多不应该提出
这个问题的负面声音
不断把它们提出来
而且他们提出的事情根本不会发生
他星期一在真相社交平台上补充说
我们已经对这些公司拥有巨大的形式和监管权利
有人在对人工智能和数据中心
进行一个病态的阴谋
而唯一对此感到高兴的就是中国
人工智能很可能会成为本月晚些时候
特朗普在华盛顿会晤中国领导人习近平时
讨论的话题之一
中国外交部发言人郭家坤
在星期一的新闻发布会上表示
人工智能的发展应以开放和包容的方式进行
他说 散播各种威胁叙事
搞对抗和恶性竞争
只会干扰人工智能全球治理进程
不符合任何一方利益
美国和中国官员的评论是继
美国人工智能巨头Anthropic负责人
达里奥·阿莫迪星期六
发布的一封公开信之后做出的
阿莫迪说
如果中国在人工智能发展上领先
将对美国和全世界构成严重危险
他还提出了一系列措施
他认为这些措施既可以确保
美国在人工智能方面保持领先
也有助于缓解潜在的安全风险
他呼吁禁止向中国出售
高性能人工智能芯片和其他设备
并加强人工智能公司的安全
以防止潜在的知识产权盗窃
阿莫迪还表示
美国的人工智能公司
应该允许独立的评估人员
访问他们的系统
以监控潜在的安全风险
并与其他民主国家协调
推动安全的人工智能发展
同时保持对中国人工智能产业的优势
另一家美国人工智能公司OpenAI的负责人
萨姆·奥尔特曼对阿莫迪的提议表示欢迎
他说放缓发展并不意味着停止人工智能的进步
而且避免代价高昂的负面结果是非常值得的
以上是美国之音的报道
唐纳德·特朗普总统说
乌克兰和俄罗斯都已经同意
停止打击对方的能源目标
特朗普9月14日
在他的真相社交平台上写道
乌克兰已同意
不打击俄罗斯能源目标
俄罗斯也已同意这样做
全球柴油价格上涨
主要是由俄乌战争引起的
不是伊朗造成的
此前特朗普星期日
呼吁乌克兰停止攻击
俄罗斯的炼油厂
他对记者说
乌克兰总统弗拉基米尔泽连斯基
必须停止破坏俄罗斯的柴油供应
并说还有许多其他目标
这些袭击正在伤害世界
乌克兰表示
俄罗斯炼油厂是合法的军事目标
因为石油工业资助莫斯科的战争
乌克兰和俄罗斯尚未对特朗普
星期一发布的真相社交媒体贴文发表评论近几周来
双方都加强对对方能源和经济基础设施的打击
俄罗斯也开始袭击乌克兰与其西方盟友之间的边境口岸
北约秘书长马克·吕特批评了俄罗斯在乌克兰波兰边境发动的无人机袭击
并且表示这将促使北约增加对基辅的支持
吕特9月14日在布鲁塞尔北约总部说
这些靠近北约领土的袭击显示俄罗斯总统弗拉基米尔普京的绝望
以及他散播恐惧和恐怖的愿望
您正在收听美国之音的中文广播
我是徐湘芸
美国总统唐纳德·特朗普星期一表示
伊朗领导人希望与美国达成协议
美国对此抱持着开放的态度
来听美国之音的报道
特朗普总统在真相社交平台上发帖说
陷入困境的伊朗急切地想要达成协议
至于美国是否会选择介入将由我来决定
我们对这一构想持开放态度
特朗普总统在另一篇真相社交贴文中指出
能源供应仍在通过霍尔木斯海峡运输
他同时表示
那些依赖海湾地区石油的国家应当向美国支付费用
以补偿美军作为引导和协助通过该水道的商船
所付出的努力
特朗普总统说
石油正流经霍尔默斯海峡
当这场骗局般的冲突结束后
此前从未向我们提供任何帮助的世界各国应当
也将会向美国支出补偿
我们这样做更多的是为了他人
而非为了自己
而且这种情况已经持续了好几代了
美国和伊朗相互指责
对方违反了在巴基斯坦和卡塔尔斡旋下
于6月18号签署的临时协议
在伊朗军方袭击了三艘途经霍尔木斯海峡的船只后
特朗普总统于7月7号宣布与伊朗的停火状态结束
自那以后,谈判陷入停滞
袭击发生后仅数日
美国便恢复了对伊朗港口的海上封锁
与此同时
德黑兰方面拒绝就其核计划展开谈判
转而与阿曼就霍尔木兹海峡的管理问题进行了磋商
作为斡旋方之一
卡塔尔上个月呼吁美伊双方重回此前签署的谅解备忘录
并寻求就海峡问题达成协议
多哈方面表示
这将使我们能够重返谈判框架
以上是美国之音的报道
为期两天的新德里金砖集团峰会
9月13日落幕
主办国印度展现了协调和化解的功夫
不仅催生出中东战争交战方都能接受的峰会宣言
也守住金砖集团非西方但不反西方的定位
来听美国之音特约记者杨安的报道
第18次金砖峰会发表新德里宣言
呼吁中东地区保持最大程度的克制
避免采取可能恶化局势的行动
中东战争无疑是本次金砖国家领导人峰会
最受关注的议题
会员之后的金砖集团内部分歧也伴随而来
新德旅选员能够顺利铲除
被分析人士视为印度在外交上的重大成果
2024年成为金砖会员国的伊朗
以及阿联酋与沙特阿拉伯
在这场中东战争站在对立面
印度习夫南达大学副教授郑嘉宾
接受美国之音采访时表示
在全球局势如此严峻的情况下
印度能够把这些国家聚集在一个平台上
这件事本身就是一种成就
5月的金砖国家外长会议上
伊朗坚持谴责美伊打击其核设施与基础设施
阿联酋则要求对等谴责
伊朗对波斯湾邻国的无差别打击
最终难以产生联合宣言
台湾清华大学印度中心主任方天次
对美国之音说
印度也算是完成这个任务。
金砖国家包含反美色彩强烈的伊朗、中国、俄罗斯
以及相对与美国亲近的阿联酋和沙特。
淡江大学国际事务与战略研究所助理教授林小珍
通过书面方式向美国之音表示,
虽然中国在金砖拥有最大的结构性影响力,
却仍无法单方面决定集团方向。
新德里宣言中以相对中性的用词
表达对中东地区紧张情势的关切
同时谴责针对民用基础设施和核设施的攻击
该宣言虽被视为间接批美国和以色列
但不把话说白既为各方保留余地
也是向美国总统唐纳德·特朗普示好的信号
以上是美国之音特约记者杨安的报道
五角大楼向美国之音证实
负责中国、台湾和蒙古事务的首席主任
新西亚、赞西、卡拉斯
将代表美国战争部
出席星期二在北京开幕的香山论坛
一名战争部官员
星期一通过电子邮件告诉美国之音
我们可以确认
战争部、中国、台湾和蒙古事务首席主任
新西亚、赞西、卡拉斯
将代表战争部出席今年的香山论坛
五角大楼没有说明
卡拉斯是否计划与中国国防部长董军
或其他中国人民解放军官员举行双边会谈
路透社此前援引一名美国官员报道
卡拉斯预计将率领美国代表团与会
他的出席级别高于去年
当时美方派出美国驻华使馆的一名国防武官参加论坛
第十三届北京香山论坛定于9月15日至17日举行
主题为凝聚稳定共识 同促安全治理
中国国防部说
本届论坛将讨论国际秩序与战略稳定
亚太安全风险
海上通道安全
新兴技术军事应用
以及太空安全治理等议题
发言人蒋冰9月14日表示
本届论坛有近100个国家
地区和国际组织的官方代表团
以及专家学者参会
董军预计将在开幕式上发表讲话
这次会议举行前不久
美国太平洋司令部司令萨缪尔帕帕罗海军上将将与负责台海方向行动的解放军东部战区司令杨志斌举行首次面对面会谈
美国之音继续播送中文节目
伊朗革命卫队上周宣布在霍尔木兹海峡捕获了一艘美国无人潜艇器
美国媒体报道说
这个潜航器符合美国安杜里尔公司的无人潜航器
美军在霍尔木兹海峡使用这款潜航器
执行无人海上监视、海底测绘以及反水雷作战
而台湾也使用这款潜航器
巡检海底电缆、保护水下安全
外界担忧伊朗将从这艘潜航器当中获得数据
但是专家表示
伊朗能从这艘掳获的故障潜航器中获得
并且与中俄分享技术有限
来听报道
美军使用美国安杜里尔公司的DiveLD无人潜航器
在霍尔木兹海峡执行无人海上监视
海底测绘及反水雷作战
而这款潜航器也被使用在台湾周边海域
根据台湾媒体的报道
台湾国家中山科学研究院
今年8月接收美国安杜里尔公司的DiveLD无人潜航器系统
并首先应用于海底电缆询检能力
以防范蓄意破坏
伊朗革命威队9月8日宣布
在霍尔木兹海峡入口处捕获了一艘美国无人潜航器
并公布了该无人潜航器的画面
经比对符合美国安杜里尔公司的DiveLD无人潜航器
美国中央司令部表示
这架无人潜航器在被伊朗发现一天多前就已经发生故障
该潜航器属于旧型号
既为收集敏感数据
也为携带任何机密的声纳或雷达设备
英国皇家联合军种研究所编辑
凯文·罗兰兹教授在接受路透社访问时表示
无人潜航器制造商和美国方面
应该都已经制定了相关规程
确保一旦设备落入敌手
最关键的技术将无法被使用
对现阶段的伊朗而言
硬件或许更具价值
这些系统所采用的工程技术
超出了伊朗目前的能力范围
伊朗确实拥有大量无人机
也曾向俄罗斯提供无人机用于俄乌冲突
且擅长使用无人系统
但在推进、制导、机动性、深度控制
以及特定通信技术方面
伊朗的技术水平可能不如美方系统先进
因此通过逆向工程
他们或许能深入了解美方的技术路径
并着手进行仿制
如果伊朗选择与俄罗斯或中国分享这些技术
我们知道它们之间确实存在此类关系
那么无论是美国还是我们其他人
恐怕都无法阻止这种技术外溢到伊朗以外的地区
根据安杜里尔
Dive LD是一具大直径自主水下航行器
长5.8米 直径1.2米
能够自主执行长达10天的任务
最大前身可达6000米
外壳采用了3D打印技术来实现高度克制化能力
并针对内部核心结构部件以及福利系统
采用了创新的制造工艺
可以为适配不同类型的载荷与传感器而进行定制改装
能够应用在国防或商用业务上
安杜里尔并提供企业租赁的方案
无论短期还是长期来看
我认为这不会造成太大影响
在任何冲突中
我们都应该预料到
会有系统损失 被毁或被服
我们只需接受这一点
将其视为常规战争的一部分
并从容应对
根据检视防务周刊与台湾媒体报道
台湾中科院与安杜里尔合作
针对台湾的需求
对Dive LD进行适应性及定制化的改造
整合多波数扫描能力
以便探测水雷和潜艇
是一款高度定制化的无人潜航器
美国之音记者刘安明 华盛顿报道
星期一股市三大估值
道琼斯工业平均指数下跌了0.29%
标普500指数下跌了0.48%
而以科技股为主的纳斯达克综合指数
则是下跌了0.56%
美东时间周一到周五晚上六点
北京时间周二到周六早上六点
敬请收听美国之音中文广播
也欢迎您浏览美国之音中文网
voachinese.com
并关注我们的XYouTube
脸书等社媒平台
获取更多的新闻讯息
美国人工智能公司Anthropic说
它在过去八个月里
阻断了与七家中国人工智能实验室
有关的模型能力提取行动
其中归因于阿里巴巴的操作者
在三个月内向Claude
发起超过1.51亿次交互
来听美国之音记者林峰的报道
Anthropic 9月10日
在最新一期人工智能威胁情报报告中说
它发现与阿里巴巴、月之岸面
Deepseek和小米等中国企业有关的操作者
试图大规模获取Cloud生成的答案
推理过程和其他能力
并将这些数据用于训练自己的模型
Anthropic在报告覆盖的8个月期间
总共阻断了与7家中国人工智能实验室有关的行动
Anthropic的报告在谈及与中国相关的活动时说
与阿里巴巴有关的操作者实施了该公司
期间发现的最大规模非法蒸留行动
根据Anthropic的统计
从今年5月至7月
与阿里相关的3500多个虚假账号
向Cloud发起超过1.51亿次交互
高峰时期每天接近300万次
Anthropic指称
操作者要求Cloud展示解决问题的详细推理过程
并把获得的输出用于训练和改进阿里巴巴的通议签问系列模型
这一数字远高于Anthropic此前公布的规模
今年6月
该公司一名政策主管在致美国国会议员的信中说
阿里相关操作者从4月22号至6月5号
与Cloud进行了约2880万次未经授权的交互
Anthropy还指称
越之岸面和DeepSeek采用了一种可能引发数据隐私问题的方式
他们把用户与自家聊天机器人的实时对话转给Cloud
由Cloud生成回答
再利用这些输出训练本公司的模型
据Anthropy称
部分用户提交的内容可能包含敏感信息
相关用户可能并不知道
他们输入中国人工智能产品的问题
被发送给了一家美国公司的模型处理
Anthropic说
越智岸面相关系统向Cloud转送了约2300万次用户请求
DeepSeek相关系统则转送了约1210万次
这两项活动合计超过3500万次
如果这一指控属实
风险将不只涉及美国人工智能模型的知识产权
和商业竞争优势
也涉及用户数据
是否在未经充分告知或同意的情况下
被交给境外第三方处理
Anthropic的报告
还把中国相关风险
扩大到国家支持的监控活动
该公司说
它发现多个巨信
与中国政府有关联的账户群
使用Cloud协助搜集、整理和分析
海外、个人及团体的信息
其中一项被Anthropic
归因于中国相关行为者的行动
重点搜集身在叙利亚的维吾尔人资料
尤其关注加入新成立的叙利亚军队
或与当地武装组织有关的维吾尔人
另一组被Anthropic评估为
与中国政府有关联的账户
则使用Cloud建立亚洲宗教领袖
和宗教团体成员的个人档案
报告所列目标包括
亚洲多名资深天主教书记
台湾基督长老教会领导人
藏传佛教公民社会成员
以及西藏流亡行政体系的相关人员
报告还称
其他相关账户利用Cloud搜集海外异议人士和华侨华人社群的信息
不过,这些指控是来自Anthropic依据账户活动
网络基础设施和使用模式得出的结论
报告没有公开相关账户的真实身份
没有说明他们具体隶属于中国哪个政府部门
当前争议的重点是
一些企业是否通过虚假账户
第三方接口和代理服务
规避地域与访问限制
大规模获取专有模型输出
并违反服务条款
将其用于训练竞争产品
美国之音记者林峰
华盛顿报道
在全球范围内
监测学术自由及保护学者的组织
学者风险网络
9月9日在官网公布了一封
寄给中国国家主席习近平的电邮
敦促中国政府立即释放遭非法拘押的美籍缅甸艺学者敏辛
来听美国之音记者吴欣欣的报道
学者风险网络组织执行主任罗伯特奎恩在电邮中写道
他代表该组织就美国公民兼政治学者敏辛的处境表达关切
他说美国国务卿马尔克·卢比奥已认定敏辛在中国遭到非法拘押
且目前下落不明
学者风险网络敦促中方确保敏欣立即获得无条件释放
电邮写道 6月3日 敏欣持有效签证抵达云南昆明机场
准备硬要参加一个学术研讨会时遭到拘押
在与外界失联一个多星期后
中国当局公开指控它涉嫌间谍活动及危害国家安全
但并未披露支持该指控的可靠证据
电邮引述公开报道说 敏欣遭拘押以来
美国外交人员仅获得了极有限的领事探视机会
但同样是美国公民的敏心妻子希尔维亚和十几岁的女儿素
却无法与她通话
敏心一直被单独囚禁于一处未公开的场所
处于指定居所监视居住之下
电邮指出 现有迹象表明
敏心遭拘押是因其和平行使学术自由与言论自由权利所致
而这些权利受到包括世界人权宣言和公民权利和政治权利国际公约在内的国际人权文书的明确保护
电邮表示 明星被捕的方式 下落不明的状态以及持续的拘押
引发了我们对正当程序是否得到落实 以及它未来境遇与安危的严重关切
电邮强调 学者们不仅深切关注明星本人的处境
以关注其遭拘押仪式向美国的中国问题学者发出的信号
电邮表示 学者们坚信
包括米星及其机构 缅甸战略与政策研究所的所有学者
都应能够在不受恐吓或任意拘押威胁的情况下开展研究
电邮最后呼吁中国立即且无条件释放米星
并感谢习近平主席关注这一重要事项
这封电邮同时抄送中国外交部长王毅
中国驻美大使谢峰
美国驻华大使庞德伟
美国务卿卢比奥
以及四位美国驻联合国官员
这是学者风险网络组织
两个星期内第二次为明星公开发声
8月27日
该组织公布了一份
有100多位美国及其他国家学者和教授的联名信
对于明星赴华参加学术活动期间
比中国非法拘押表达严重关切
目前 这份公开信已经有113位学者签名表达支持
9月9日是敏昕的生日
敏昕的弟弟昂茂新8月24日在接受美国之音专访时说
我们真的很希望我哥哥能在他生日前获得自由
9月8日 敏昕创建的机构缅甸战略与政策研究所
在其官网发布一篇题为《他不在的生日》的评论文章
披露明星每逢生日时
通常都会回顾自己的人生历程
并在社媒上发布感言
分享内心的深刻思考
以上是美国军营记者吴欣欣在华盛顿的报道
英国军情武处处长肯·麦卡勒姆
在星期五美国911恐怖袭击事件
25周年当天撰文警告
英国在大规模恐怖袭击风险之外
还面对实力强大的国家
构成的战略性安全威胁
来听美国之音特约记者李伯安
从伦敦发来的报道
就在不久前
7月份刚刚就任的英国首相安迪·伯纳姆
刚与中国国家主席习近平首次通话
麦卡勒姆在《独立报》发表的文章中
虽然没有点名任何国家
但他此前对中国威胁的界定
以及英国今年审节的首宗侍华间谍案
令新政府如何兼顾对华经济接触
与国家安全在受审视
据唐宁街说法
伯纳姆在与习近平的通话中
讨论了促进经济增长的务实合作
也提出了国内安全、香港和人权问题
麦卡勒姆在2025年10月的年度威胁报告中
曾经表示,过去一年,英国军情五处针对涉嫌参与国家威胁活动而进行调查的人数增加了35%
他列出的中国相关活动包括网络间谍、秘密转移技术、暗中干预公共生活以及骚扰和恐吓意义人士
他同时主张英国应该抵御威胁
也应把握符合国家利益的合作机会
今年6月份
五眼联盟安全机构联合警告
中国军事情报机关正利用职业社交和招聘网站
接触可获取敏感资料的公务员、军人、记者和智库人员
不过英国军情五处没有公布涉华案件的比例
也没有披露对华工作的人员和预算
英中透明组织主任萨姆·邓宁对美国之音说
如果威胁确如他所说正在上升
那么资源是否也在增加
现有资源是否足够
被香港警方悬赏通缉
现居英国的民主活动人士郑文杰说
跨国镇压最深的影响在于
疑似监控、网络骚扰和向香港亲属施压所形成的寒蝉效应。
他说,一个外国政府不必成功监控每一个人,
只要有足够多人相信自己可能被监控、恐吓,
就已经实现了部分目的。
他认为,中央政府对威胁的认识尚未转化为地方警察
对受害者一致而有效的支援
多次代理香港和中国异议人士庇护
以及红色通报案件的大律师
本基斯对美国之音说
从其经手的案件来看
中国相关行动会同时动用
网络技术、人力情报和外交资源他说
他们愿意投入大量的资源追踪一些
实际上并不构成重大威胁的人
换言之
目的往往不是窃取机密
而是监控和压制
可能令北京难堪的人
以上就是美国之音中文广播的全部内容
感谢您的收听
您也可以到美国之音中文部的YouTube频道
或者到脸书等社媒平台
收听美国之音中文广播
您也可以在苹果播客和Spotify
搜索美国之音随时收听我们的节目
我是许湘芸
我们下次节目时间再见
Replies (0)

No replies yet — be the first to comment

Post a reply