China's new national standard takes aim at the 'transfer me to a human' runaround
China's first national standard governing how AI and human customer service work together takes effect September 1 — and it goes straight at the most hated feature of modern support: the bot that won't hand you to a person. Under GB/T 47746—2026, "Customer contact service — Requirements for collaboration between human and intelligent customer service," human-service entry points must be clearly visible rather than buried in menus, AI handles routine questions while complex disputes get escalated promptly, customers shouldn't have to repeat themselves after a bot-to-human switch, and issues touching life or property safety must transfer to a person automatically.
The timing is no accident. Consumer associations logged roughly 985,900 complaints nationwide in the first half of 2026, with after-sales service the largest category at 26.79% of the total — and hard-to-reach humans, unfulfilled AI promises, and wrong answers cited as the new flashpoints. The horror stories write themselves: one seller's platform-hosted AI agent knocked a ¥1,000 sound card down to ¥400 on its own, while other users have burned through a dozen "转人工" ("transfer me to a human") messages only to receive looping "I understand your feelings" replies. Newsroom tests this week found only 4 of 20 major platforms connected a human quickly, averaging 171 seconds, with the worst case dragging past 11 minutes.
Our take: this is the first consumer-grade regulation aimed not at what AI says but at how it's wired into an org chart — and other countries watching the same frustration build should take notes. Compliance will be cheap; the reputational math of deliberately using bots as complaint-absorption armor just changed. Enforcement teeth are the open question: a recommended national standard (GB/T) sets expectations for platforms and gives regulators and courts a reference, but it isn't a licensing regime.
What to watch: whether e-commerce and delivery platforms quietly rework their escalation flows before September 1 — and who gets named-and-shamed first.
Have you ever begged an AI support bot for a human without success?
RayNeo's new iO glasses drop the camera and bet everything on text overlays. While every rival stuffs cameras and speakers into eyewear, RayNeo's launch leans the other way: no camera, no speaker, just a green waveguide display at 97% transparency and about 1,300 nits pushing captions, summaries, and teleprompter lines into your view. Four mics plus a bone-conduction sensor let it listen to conversations, extract action items, and accept calendar suggestions with a head nod, with speech translation across 40 languages. Out of the box it runs Gemini 3.1 Flash Lite; $9.99/month unlocks ChatGPT 5.1, Gemini 2.5 Pro, Claude, and DeepSeek — several already superseded, which says a lot about how fast model churn outruns hardware roadmaps. A privacy-first angle in a camera-skeptical market is a genuinely interesting bet.
Sources: Sina Finance · CCTV/央视新闻 · IT之家 · 红星资本局 · The Decoder · RayNeo
