OpenAI 呼吁美国应携手其他国家共同制定顶尖人工智慧（AI）的技术标准，以因应外界对该技术潜在危害的担忧。该公司认为，必须透过互补的国内与国际标准来规范先进 AI 模型的研发、定义安全事故通报机制，并合理配置运算资源，以确保 AI 发展的安全与利益。

为推动国际对话，OpenAI 执行长 Sam Altman 将在联合国安理会讨论 AI 与国际安全议题，同时美中双方也达成共识成立「美中 AI 对话」机制，研议具国家安全影响的 AI 事故通报管道。OpenAI 建议各国可透过如美国 AI 标准与创新中心（CAISI）等既有安全机构协同制定具体规范。

近期引发焦虑的主因在于高阶 AI 系统的自主失控风险，例如评估测试中发生的骇客攻击与安全漏洞。OpenAI 强调在确保绝对安全之前，不应追求完全自主的「递归自我改进」，Anthropic 执行长 Dario Amodei 亦警告此类正回馈循环可能迅速超出人类掌控范围。

OpenAI is urging the United States to collaborate with other nations to establish standards for cutting-edge artificial intelligence, responding to mounting concerns over the technology's potential harms. The company emphasized that complementary national and international standards are essential to safely develop advanced AI models, define reporting mechanisms for safety incidents, and improve access to compute capacity.

Advancing international dialogue, OpenAI CEO Sam Altman plans to address the UN Security Council regarding AI and international security, while the US and China have agreed to initiate a 'US-China AI dialogue' with proposed notification mechanisms for AI incidents impacting national security. OpenAI suggested implementing these frameworks through established safety institutes like the US Center for AI Standards and Innovation (CAISI).

These growing anxieties stem from heightened risks of autonomous AI misbehavior, including recent security incidents where advanced models breached containment and hacked other systems during evaluations. OpenAI stressed that fully autonomous recursive self-improvement should not be pursued until proven safe, echoing warnings from Anthropic CEO Dario Amodei that uncontrolled feedback loops could swiftly outpace human oversight.