Beijing Watch: Did Apple Siri Quietly Connect with Alibaba's Qwen? Chinese Website Briefly Reveals AI Document Then Pulls It, Cyberspace Administration Responds!

On August 8, Apple's official China website unexpectedly published a support document titled 'Using Qwen with Apple Intelligence on Mac' in its Mac user manual. The document clearly stated that mainland users with macOS 26.6 or later could access generative AI capabilities from Alibaba's Tongyi Qwen (Qwen) through Siri AI and writing tools. Examples included photo and document analysis, PDF summarization, and directly telling Siri, 'Please ask Qwen to write a poem about dragons.' This was interpreted by outsiders as a key signal that Apple's China-specific version of Apple Intelligence was officially rolling out.

However, within 24 hours, by August 9, the document was quietly removed, and the page reverted to the previous version titled 'Using ChatGPT with Apple Intelligence.' The original link is now inaccessible. Apple customer service responded that the retraction means the feature is temporarily unavailable. They further stated that the service has not yet officially launched in mainland China and remains under regulatory review, with the official launch date subject to further announcements.

More importantly, this integration is not simply about users 'opening the Qwen app'—it embeds Qwen directly into Siri and Apple's native tools. At WWDC26, Apple officially unveiled its next-generation Apple Intelligence and a revamped Siri AI, emphasizing that the new Siri would have enhanced contextual understanding, screen awareness, personal information comprehension, and broader knowledge capabilities. Apple also explicitly stated that until relevant regulatory requirements are met, the new Siri AI and other Apple Intelligence features would not be available in mainland China.

In May, China's Cyberspace Administration (CAC) released updated generative AI registration information for March–April 2026, reiterating that generative AI applications or features that directly call upon registered models via API or other means must also complete the required registration.

On July 15, the government officially released its first list of 'on-device generative AI services for smartphones,' with 'Apple Intelligence' formally appearing on the list. On the same day, Alibaba confirmed to media that Qwen would be integrated into Apple Intelligence to serve iOS, iPadOS, macOS, and visionOS users in the Chinese market, providing capabilities in text understanding, image understanding, and content generation.

From the leaked operational guidelines, the most significant aspect of this integration is not merely that Apple is using Qwen, but where Qwen is being placed. Previously, using large language models like ChatGPT or Qwen required users to actively open an app, website, or chat interface. Apple Intelligence, however, operates on the opposite logic: AI should be an integral part of the operating system.

There is currently no public evidence suggesting Apple and Alibaba have terminated their collaboration on Apple Intelligence and Qwen. On the contrary, the official registration published in July and Alibaba's public confirmation indicate that the partnership is not mere market speculation.

Alibaba Group owns Alibaba Cloud, and the Qwen app is fully integrated into Alibaba's ecosystem, including Alipay, Taobao Instant Purchase, Fliggy, and Gaode Map.

Why does Apple need Qwen in China—and not another model? The reporter contacted China's Cyberspace Administration, where an official stated, 'The relevant rollout work is ongoing, but it is still not available for use.' Regarding the timeline, the official said, 'It will only be possible after regulatory review is passed.'

Apple has consistently aimed to maintain control over the core Apple Intelligence experience. However, in the Chinese market, AI models cannot simply go live based on technical superiority alone. Model services involve data governance, content moderation, generative AI registration, and compliance with regulatory requirements.

In June, Apple publicly stated that Apple Intelligence and Siri AI would only be provided in mainland China after meeting relevant legal requirements. Partnering with a local large-model provider like Alibaba may be one of the most realistic solutions.

Previously, Alibaba's Qwen primarily reached users through its own app, website, and cloud services. Once integrated into Apple Intelligence, Qwen no longer requires users to actively open the Qwen app. It can directly appear within Siri, writing tools, photos, documents, and other native Apple experiences.

Privacy and other concerns remain important issues for Chinese tech firms. Apple Intelligence is fundamentally a product and system framework, with large language models serving as just one source of capability. What Apple truly controls are the user interface, system permissions, data orchestration, hardware capabilities, and the interaction between users and AI. This may explain why Apple is adopting a 'partner model + Apple Intelligence framework' approach in the Chinese market.

For Apple, this is a relatively pragmatic compromise: the model is localized, but the AI gateway remains under Apple's control. When Siri begins saying, 'Let Qwen help,' it reflects a complex network of regulation, business, and geopolitics. Whether the feature will officially launch in the next system update or undergo further adjustments remains to be seen, pending further actions from Apple and regulatory authorities.