A wave of eerily identical incorrect default addresses is unsettling numerous Taobao shoppers. Recently, a surge of complaints has appeared on social media, with many users discovering their default delivery addresses had been silently changed to "National Museum of China, Li Chun," resulting in misdirected orders, lost packages, and a cascade of after-sales disputes.
This "ghost address" phenomenon is not an isolated incident. Around the same period, many consumers have also voiced frustration over sudden account bans, attributing them to an overzealous AI risk-control system that misidentifies routine shopping behavior as "web crawling." Adding to the bewilderment, some shoppers have reported seeing an AI-generated advertisement on Taobao's homepage featuring a model with three arms, raising questions about how such obviously flawed content made it to the platform's prime real estate.
The mystery of "Li Chun" has become a central puzzle. Numerous users report that their default shipping information was modified without their consent, with all instances uniformly pointing to "National Museum of China, Li Chun." This has led to significant numbers of orders being sent to the wrong destination, with some consumers losing goods entirely and others facing complex return or exchange procedures. One frustrated user described paying for an order in the morning only to notice the incorrect address at the last moment, a change they could not trace or explain. Contacting customer service yielded little, with only a suggestion to change their password and no clarity on how or when the address was added. The phone number associated with the address appears to be either disconnected or answered by strangers who deny being "Li Chun," suggesting the recipient is not a real person.
The sense of a digital "groundhog day" is palpable. Even after users manually delete the erroneous address, reset their defaults, or change their passwords, the "National Museum of China, Li Chun" entry often reappears after a while, seemingly immune to correction. This has fueled speculation among users, with theories ranging from a backend data error at Taobao, to merchants exploiting platform vulnerabilities for fraudulent orders or shipping insurance scams, to deeper concerns about a systemic leak of user data. When contacted for comment, a staff member at the National Museum of China stated they could not verify individual personnel information and suggested directing inquiries to Taobao itself.
Experts point to a combination of vulnerabilities. Guo Tao, an angel investor and senior AI specialist, explains that the mass appearance of uniformly tampered ghost addresses points to a dual weakness in batch account data theft and the platform's address storage verification. He notes a lack of strong secondary verification for address changes, potential interface vulnerabilities that can be exploited via third-party plugins or shared login environments, and an absence of checks to prevent large numbers of accounts from being updated with the same address in a short period. Furthermore, the system seems to lack audit logs for address changes, making it impossible for users to trace the source of the modification, exposing fundamental shortcomings in data privacy and abnormal behavior interception.
Beyond the ghost addresses, the wave of account suspensions has sparked significant anger and confusion. Users report being banned for simple actions like browsing products for a few minutes, with customer service later confirming the accounts were flagged as "suspected crawlers" by the AI system. Some users describe a frustrating cycle of being banned, appealing, getting unbanned, and then being banned again without a clear explanation. This has led to a palpable sense of unease, with some long-time, high-value members feeling their loyalty is not being valued. One user, a long-standing 88VIP member with significant spending, expressed disbelief at being locked out for behavior as innocuous as browsing, highlighting a perceived flaw in the AI's judgment.
Guo Tao attributes these erroneous bans to structural flaws in the risk-control model, including an over-reliance on superficial behavioral features like high-frequency browsing, an imbalanced training dataset that lacks sufficient examples of diverse, legitimate user habits, and a static threshold system that doesn't account for user history. He also criticizes the lack of a manual review process before an account is banned, leaving a single AI model's output to determine user access without a buffer for human verification.
The platform's content review processes have also come under fire. Consumers have pointed out an AI-generated advertisement on the Taobao homepage showing a model with three hands—a basic anatomical error that should have been easily caught. While the merchant involved stated they had notified their operations team, and Taobao's official customer service acknowledged the issue and promised a review by their commerce management department, the incident underscores a gap in quality control for AI-generated content.
In response to the broader concerns, a Taobao spokesperson highlighted the platform's ongoing efforts to combat AI-generated false images, stating that they have intercepted over 500,000 such images and rejected more than 10,000 AI-generated images used in IP complaints since March 2025. They also mentioned the introduction of an AI model to identify fake after-sale images and a mutual trust mechanism for returns to counter AI-enabled fraud. However, these measures do little to quell the immediate frustrations of users facing bans and phantom addresses.
Industry observers see these issues as a symptom of a larger problem: the pace of AI adoption is outstripping the development of corresponding governance and risk-control measures. Gao Pan, founder of Jiangnanbei Group, notes that while Taobao's intent to reduce costs and improve services through AI is correct, the tools are being released before they are fully refined. He points to the lack of proper validation for AI-generated content and the insufficient capability of AI customer service to handle complex consumer issues as key failures. He emphasizes that this is an industry-wide challenge, not unique to Taobao, where the drive for efficiency often overshadows the need for robust compliance rules and reliable human fallback mechanisms. The ultimate goal for e-commerce AI, he argues, should be to support, not replace, human judgment, ensuring that technological efficiency never erodes consumer trust.
As Taobao continues to integrate AI more deeply, including the recent integration with Qwen to power shopping assistants and features like AI try-ons and price comparisons, the question remains whether this rapid transformation is leading to greater efficiency or a loss of control. With nearly 1.55 million complaints filed on the Heitao platform under "Taobao," many of which now involve AI-related issues like ineffective customer service and disruptive features, the path forward requires a recalibration where governance, review processes, and human oversight catch up with the speed of AI innovation.
Comments