Apple's AI initiative in China faces stringent requirements before it can be publicly launched. The regulatory framework mandates that AI models respond correctly to a set of 2,000 questions, with a crucial criterion that they must refuse to answer at least 95% of queries that could reveal information deemed sensitive by the government.
As part of this compliance, Apple will utilize a partnership with Alibaba for its AI product, which includes the Qwen3 chatbot optimized for Apple devices. This move comes as foreign companies are compelled to employ local, state-approved models to operate within the heavily regulated Chinese market. The government employs rigorous testing to ensure that these models align with its censorship policies.
Moreover, specialized agencies have emerged to assist AI companies in navigating these complex requirements, akin to preparatory services for standardized tests. This environment reflects the broader landscape of online information control in China, where many western platforms are restricted and local search engines like Baidu face similar limitations.
Overall, the measures highlight the tension between technological innovation and governmental oversight, as companies strive to adhere to strict regulations while delivering effective AI solutions.