As large models rush into education and healthcare — "asking AI first for any illness" and students answering with tablets — a wave of new rules hit the brakes, The Beijing News reported on October 28. Beijing's education commission issued a 2026 AI-application guide for schools with grade-by-grade "red lines"; over 40 authoritative institutions released an expert consensus on medical-institution AI use emphasizing "human-machine collaboration, human-led"; and the National Medical Products Administration's draft guidance states AI "may only assist physicians' decisions and shall not serve alone as a basis for diagnosis or treatment."
The tightening is not about the technology itself but about posture in high-threshold, livelihood-critical scenarios. At iFlytek's 1024 festival, chairman Liu Qingfeng released the domestically trained Spark X2.6, increasingly landing in education, healthcare and law — answering the question of "how AI should be present" with: AI should be by humans' side, not in humans' place.
iFlytek's "Starlight" teacher super-agent is positioned not as a chatbot but as an assistant that "understands the teacher," integrating lesson prep, research and home-school communication. The company says it has saved 60%-80% of transactional time for 300,000 registered teachers.
The episode shows China prototyping a governance model where AI offloads repetitive work while humans retain judgment, safety and accountability.
Source: The Beijing News / iFlytek
