During this year's Double 11 shopping festival, many consumers browsing Taobao late at night frequently encountered AI streamers hawking products. These streamers have regular facial features but stiff eyes and looping scripts, and they are unable to accurately answer detailed questions raised by consumers. In addition, the AI synthesis disclosure labels are extremely inconspicuous, making them hard for most consumers to notice. This has caused widespread discomfort among users, with many saying they are unwilling to click on such livestream rooms and some even feeling deceived. Taobao is vigorously promoting the AI streamer ecosystem from the top down, launching paid digital human services and supporting them with public-domain traffic support policies to attract merchants to enter in batches, in an attempt to use AI to restructure e-commerce and tap new growth points. However, at present, most leading livestreaming MCN institutions have rarely entered, and many peers who previously tried AI streamer projects have also put their plans on hold due to poor results and difficulty in building user trust. By contrast, platforms such as Douyin are stricter in regulating purely unattended, looped pre-recorded AI sales livestreams, out of concern that they may involve violations such as false advertising. Experts point out that currently mass-commercialized AI digital human solutions are limited by technology and lack details such as micro-expressions and vocal emotion, with the sense of incongruity amplified in late-night scenarios. At this stage, AI streamers are only suitable as a supplementary tool during low-traffic periods, handling standardized explanation work, and cannot complete core links such as trust building and complex communication. Human-machine collaboration is the reasonable implementation model at present. Taobao has previously been repeatedly exposed for chaos in AI-generated content, including deformed limbs in AI model images and failure to label AI-generated attributes as required. Its AI content governance overall leans toward after-the-fact remediation, and its pre-emptive prevention and control capabilities have shortcomings. Problems such as the heavy pressure of verifying massive amounts of material, merchants deliberately concealing AI attributes, and the difficulty of tracing responsibility for violations have increased the difficulty of long-term platform governance. If the chaos of low-quality and opaque AI streamers is not constrained in a timely manner, the original intention of reducing costs through technology may continue to erode consumers' trust in the platform.