美国政府审查后OpenAI发布新模型，AI巨头竞相上市与办公场景争夺战升温
在经历美国政府安全审查后，OpenAI推出GPT-5.6系列及工作代理工具，与Anthropic的IPO竞赛和二级市场狂热同步展开，企业客户成本压力与版权诉讼增添变数。
OpenAI于本周四正式向公众推出其新一代GPT-5.6模型系列，包括旗舰模型Sol、中端版本Terra和低成本选项Luna。此次发布此前因美国政府要求而推迟，华盛顿方面对前沿AI模型的网络安全能力表达了担忧，并进行了数周的技术测试与红队演练。OpenAI首席执行官萨姆·奥尔特曼称，公司在审查过程中根据政府反馈做出了“许多修改”，并形容这一互动为“协作性的来回”。该事件标志着美国政府对最先进AI模型部署的介入正从临时安排走向制度化，白宫正在制定行政令框架，以确定哪些模型需接受强制性安全审查。
与此同时，OpenAI与其主要竞争对手Anthropic均在加速推进首次公开募股（IPO）计划，两家公司已秘密提交申请，目标估值均接近万亿美元。二级市场数据显示，Anthropic的估值已飙升至1.2万亿美元，较5月融资轮上涨近25%，但股票几乎无人出售，需求远超供给，催生了大量未经授权的特殊目的载体（SPV）交易。Anthropic已明确警告投资者警惕此类交易。OpenAI的估值则回升至约9080亿美元，其新模型发布重新点燃了买方兴趣。两家公司采取了不同路径：OpenAI凭借ChatGPT的庞大用户基础主攻大众市场，而Anthropic则凭借更高的经常性收入预测吸引机构投资者。
在产品层面，OpenAI将代码工具Codex整合进ChatGPT桌面应用，并推出“ChatGPT Work”代理，可连续数小时执行跨应用任务，生成文档、表格和演示文稿。此举旨在将ChatGPT打造为职场一站式平台，直接回应企业客户对降本增效的强烈诉求。奥尔特曼承认，企业AI支出已成为今年最受关注的议题，客户要求更快的速度和更低的成本。Anthropic此前也发布了类似产品Claude Cowork，双方在办公自动化领域的竞争进一步白热化。
法律风险同样在积聚。以《纽约时报》为首的报业集团向曼哈顿联邦法院申请对OpenAI实施制裁，指控该公司在版权诉讼中谎称无法搜索其模型中的受版权保护内容，而实际上已进行过此类搜索，并删除了数十亿条相关聊天记录。这一争端可能影响生成式AI训练数据的合法性边界，为整个行业带来深远影响。
未来数月，两大AI企业的IPO进程将是关键观察点。若Anthropic按计划于今秋上市，其市场表现将为OpenAI后续的上市窗口定下基调。同时，美国政府即将出台的AI模型安全审查细则，将决定下一轮模型发布是否会再次遭遇延迟。企业客户对AI投入回报的审视，也将持续考验两家公司的定价能力与产品价值。
| 印度及南亚媒体 | −0.40 | critical |
|---|---|---|
| 大西洋／英语圈媒体 | 0.00 | neutral |
| 伊朗及相关媒体 | +0.60 | aligned |
| 东南亚媒体 | +0.50 | aligned |
US regulation, criticized as anti-innovation, is inadvertently accelerating the adoption of open-source models, while American giants continue to launch new models.
The bloc presents regulation as a mistake that backfires, using the term 'anti-regulation White House' to delegitimize restrictions and suggest open source is the answer.
It does not mention the national security concerns that motivated the regulation, nor the additional testing process that allowed the launch.
National security is the priority: the launch of GPT-5.6 was approved only after additional testing, and experts warn of cyberattack risks.
The bloc adopts a hierarchy of threats, placing national security at the center and legitimizing government intervention as necessary, citing experts to strengthen credibility.
It does not mention the open-source surge or Chinese models as a consequence of regulation, nor the SpaceXAI launch.
OpenAI has launched a revolutionary new voice assistant that makes conversations with AI more natural than ever.
The bloc depoliticizes the news by focusing exclusively on technical features and user experience, completely ignoring the regulatory context and controversies.
It completely ignores US regulation, the open-source debate, and the SpaceXAI launch, presenting a different event as if it were the main news.
OpenAI launches GPT-Live, a model that makes conversations with AI so realistic they feel human.
Similar to the Iranian bloc, it depoliticizes the news by emphasizing technological innovation and user experience, with no reference to regulatory issues.
Same omission as the Iranian bloc: no mention of US regulation, open source, or SpaceXAI.