白宫与Anthropic举行“建设性”会谈，讨论Mythos模型安全风险White House and Anthropic Hold 'Productive' Meeting Over Mythos AI Model Risks

Anthropic于上周发布了Claude Mythos预览版，这一举动立即引起了华盛顿监管机构和安全专家的警觉。该公司宣称，Mythos在处理复杂的网络安全任务时具有极高的效率，甚至在某些特定领域能够超越人类专家的表现。这种能力虽然在防御端具有巨大价值，但也引发了关于该技术可能被滥用于高科技犯罪或国家间网络对抗的担忧。白宫官员希望通过此次会谈，深入了解该模型的内部机制及其可能带来的系统性风险。

The release of the Claude Mythos preview last week immediately triggered alarms among Washington regulators and security experts. Anthropic has claimed that the model possesses unprecedented efficiency in handling complex cybersecurity tasks, even outperforming human experts in specific domains. While such capabilities offer significant defensive value, they also raise concerns about the technology being weaponized for high-tech crime or state-sponsored cyber warfare. White House officials sought the meeting to better understand the model's internal mechanisms and the systemic risks it might pose to national infrastructure.

尽管白宫幕僚长与Anthropic进行了深入交流，但特朗普总统本人似乎对此次具体会晤并不知情。在亚利桑那州凤凰城的一次采访中，当被问及阿莫代的访问时，特朗普总统反问道：“谁？”并表示他对此“一无所知”。这种反应突显了白宫内部在处理AI政策时，行政幕僚的专业化运作与总统公开表态之间的微妙差异。尽管总统此前曾对该公司持强硬立场，但幕僚层显然认为与Anthropic的对话对于掌握前沿AI动态至关重要。

Despite the deep engagement between the Chief of Staff and Anthropic, President Trump appeared personally unaware of the specific meeting. During an interview on a runway in Phoenix, Arizona, when asked about Amodei's visit, President Trump responded, "Who?" and stated he had "no idea." This reaction highlights a potential disconnect or a strategic division of labor between the administration's professional staff and the President's public stance. While the President previously enforced a hardline policy against the firm, his senior advisors clearly view direct dialogue with Anthropic as essential for monitoring cutting-edge AI developments.

此次会议的背景极为复杂，Anthropic目前正因合同或政策争议起诉美国国防部。与此同时，特朗普政府此前对该公司的强硬立场——包括将其列入黑名单——反映了政府对AI公司自主权与国家安全之间冲突的担忧。然而，Mythos模型的出现似乎证明了该技术对于美国政府而言可能过于关键，以至于无法完全将其排除在对话之外。这种“边打边谈”的局面反映了当前AI治理中技术依赖与安全监管之间的复杂博弈。

The context of the meeting is further complicated by the fact that Anthropic is currently suing the U.S. Department of Defense over contractual or policy disputes. Simultaneously, the Trump administration's previous hardline stance—including the blacklisting of Claude products—reflects ongoing friction between corporate autonomy in AI development and national security mandates. However, the emergence of the Mythos model suggests that the technology may be too critical for the U.S. government to ignore entirely. This dynamic of simultaneous litigation and high-level consultation underscores the complex balancing act between technological reliance and regulatory oversight in the AI era.