造梦者

奥特曼凤凰网科技讯 北京时间8月24日，据《商业内幕》报道，OpenAI CEO萨姆·奥特曼(Sam Altman)表示，他担心有朝一日AI会被少数几家公司、模型或个人所掌控，使得绝大多数消费者在技术如何塑造世界方面失去话语权。    (ECNS) -- China on Tuesday rejected accusations from the United States and other countries over its submarine-launched intercontinental ballistic missile (ICBM) test conducted in July at the Conference on Disarmament in Geneva.     Li Chijiang, China’s ambassador for disarmament affairs, said China opposes Washington using the forum for political manipulation and criticized the United States for questioning China’s advance notification of the missile test.        Li made the remarks at a meeting on Tuesday, responding to comments by U.S. official Darryl Nirenberg, who accused China of not providing pre-launch notification of the July 6 test to the countries regarding the submarine-launched ICBM test.    Li said the U.S. had cited the voluntary political commitment of the Hague Code of Conduct Against Ballistic Missile Proliferation (HCOC) to demand that China provide advance notifications of missile launches. However, China is not a member of the HCOC and is therefore not bound by its provisions.    Li also noted that China has conducted only three ICBM tests to date, the fewest among the five nuclear-weapon states. In contrast, the U.S. has conducted the most ICBM tests worldwide and continues to expand its nuclear capabilities.    China has conducted missile tests in a safe, responsible and transparent manner, he said.    Li stressed that China has always followed a self-defensive nuclear strategy, maintained its nuclear capabilities at the minimum level required for national security and does not participate in any nuclear arms race.    He urged the U.S. to reflect on its own actions and take measures that genuinely contribute to global peace and stability.    (By Helen Mo, intern Liu Shuangjing)                    。奥特曼在周日播出的戴维·森拉(David Senra)播客节目中表示：“正确的做法是，让人们牢牢掌控未来。社会和模型将共同演化。”奥特曼表示，他担心的另一种可能性是AI摆脱人类控制者的束缚。矛盾的是，他认为，对后一种可能性的恐惧可能会导致前一种可能性成为现实。“有很多人对于这些风险的严重程度感到非常紧张，并被这种担忧深深影响，并觉得自己有必要保护世界，”他说，“比如，他们会认为：我们应该牺牲大量自由，以换取安全。

B | ”奥特曼很可能是在暗指Anthropic CEO达里奥·阿莫代伊(Dario Amodei)。

C | 阿莫代伊曾在播客、文章和公司博客中多次警告AI快速发展所带来的危险。他在6月的一篇文章中写道，该公司的新Mythos模型对网络安全、金融行业、关键基础设施和国家安全构成了“非常现实的风险”。阿莫代伊利用这些警告，呼吁政府在AI监管中发挥更大的作用。在奥特曼看来，这可能是在“用自由换取安全”。

D | Anthropic也是上个月唯一一家未签署支持开放权重模型公开信的主流AI公司。开放权重模型允许开发者在自己基础设施上共享、下载、修改和部署AI模型。中国已积极拥抱开放权重模型，这使得其能够以极低的成本跟上硅谷的发展步伐。

E | 美国国内的一些力量，包括英伟达、Meta和微软，则认为美国也需要采取同样的做法，否则就有可能在AI竞赛中将阵地拱手让给竞争对手。尽管OpenAI尚未发布其前沿模型的源代码，但该公司去年确实发布了两款开放权重模型：gpt-oss-120b和gpt-oss-20b。奥特曼还在播客中表示，美国的AI领导者们未能向公众展示这项技术如何让人们对自己的生活拥有更大的自主权。“我们即将迎来有史以来规模最大的创业热潮，届时会有更多人创办小型企业，”他说，“我认为AI正在赋予人们这种能力。”(作者/箫雨)更多一手新闻，欢迎下载凤凰新闻客户端订阅凤凰网科技。

F | 想看深度报道，请微信搜索“凤凰网科技”。