OpenAI和Anthropic两家美国AI公司正积极支持澳大利亚新的人工智能监管框架 [1]。这一立场背后隐含着这两家公司面临的多重压力和战略目标。
中国初创企业Moonshot AI推出的K3模型构成了对两公司的直接竞争威胁 [1]。自该模型推出后，两家公司的市场估值大幅下滑：Anthropic的估值下跌2320亿美元至1.56万亿美元，OpenAI的估值则下跌1600亿美元至1.16万亿美元 [1]。
支持监管框架的战略意义在于多个层面。澳大利亚政府的新规将赋予创意工作者选择权，让其决定AI模型是否可用于训练其作品，并有权获得赔偿 [1]。Anthropic在这一政策环境下也正在采取行动，已与作者和解版权诉讼，支付15亿美元赔偿，其中约300万美元用于约50万本书的补偿 [1]。
这一监管立场同时服务于两公司的融资和上市计划。Anthropic已开始注册在美国股市上市，对标SpaceX在6月上市后融资860亿美元、达到2.1万亿美元估值的成就 [1]。通过支持规范化的监管框架，两公司能够为IPO打造更有说服力的商业叙事，增强投资者信心。
Anthropic CEO Dario Amodei在6月发表的论文中呼吁建立由民主国家联盟主导的AI治理体系，并建议政府对AI模型进行测试和必要的封禁 [1]。这一立场表明该公司将监管接纳视为长期战略的核心部分。值得注意的是，美国政府已强制Anthropic暂停向外国人提供Mythos 5模型的访问权限 [1]，反映出政府在AI出口管制方面的介入。
OpenAI and Anthropic are supporting Australia's new artificial intelligence regulatory framework, a position that reflects broader strategic calculations for the two American companies. [1]
The backing comes as both firms face intensifying competitive pressure. Chinese startup Moonshot AI unveiled its K3 model on Friday, positioning it as a competitor to OpenAI's ChatGPT and Anthropic's Claude. [1] The competitive challenge has coincided with significant valuation declines for both companies: Anthropic's valuation dropped 2.32 trillion dollars to 1.56 trillion dollars between Friday and Tuesday, while OpenAI's fell 1.6 trillion dollars to 1.16 trillion dollars, according to IG Markets data. [1]
Beyond immediate market pressures, the companies' regulatory support appears tied to their strategic objectives. Anthropic has begun registering for a public listing on U.S. stock markets, looking to replicate the model of SpaceX, which raised 86 billion dollars following its June market debut and achieved a 2.1 trillion dollar valuation. [1] Supporting favorable regulatory frameworks helps construct a compelling narrative for potential investors and demonstrates corporate responsibility ahead of potential public offerings. [1]
The Australian regulations themselves address concerns about AI model training. Assistant Technology Minister Andrew Charlton stated that the new rules will allow creative workers to choose whether AI models can be trained on their work and to receive compensation if so. [1]
This regulatory positioning aligns with public statements from Anthropic leadership. In a June paper, CEO Dario Amodei called for a coalition of democratic nations to lead AI governance and suggested that governments conduct testing and potential bans on AI models. [1]
Recent enforcement actions underscore the regulatory environment's tightening. The U.S. government has compelled Anthropic to suspend access to its Mythos 5 model for foreign users. [1] Additionally, Anthropic settled a copyright lawsuit with authors on Monday, paying 1.5 billion dollars, with approximately 3 million dollars allocated as compensation for roughly 500,000 books. [1]