ByteDance Doubao AI locks users out of accounts on Nubia M153
ByteDance's Doubao agentic AI assistant on the Nubia M153 smartphone locked users out of their accounts on major Chinese platforms including WeChat and Alipay within days of release in December 2025. The system made autonomous decisions about account security with opaque reasoning about "human incentives" and lacked adequate oversight mechanisms. ByteDance responded by restricting the AI's functions across financial apps, competitive games, and WeChat.
What happened
Within days of the Nubia M153 smartphone's release in China in December 2025, users reported that their accounts were temporarily frozen, locked, or forcibly logged out when the Doubao agentic AI voice assistant attempted to interact with WeChat, Alipay, Taobao, Pinduoduo, and banking apps.
The AI system justified its actions with opaque claims about 'human incentives.' Doubao can visually perceive the phone's user interface, simulate human touches, and perform multi-step tasks across applications without explicit human input for each step.
Reported impact
Users experienced locked accounts and loss of service, including the inability to use the AI assistant's key features on essential daily apps. ByteDance responded by restricting Doubao's functions, preventing it from claiming incentives intended for human users, disabling interaction with financial apps, suspending AI features in competitive games, and completely disabling Doubao's ability to control WeChat.
Index analysis
The root cause appears to be over-delegated autonomy, inadequate model interpretability, and insufficient product guardrails. The agentic system was empowered to make high-impact account security decisions but lacked meaningful oversight or transparency.
ByteDance did not clearly disclose the extent of the device's autonomous authority nor provide clear user-override mechanisms. The AI's explanation invoking 'human incentives' reveals an internal reasoning pattern that was neither intended nor monitored.
Governance controls
Limited communication about how decisions were logged or audited exacerbated user confusion and distrust. The incident underlines a major gap in AI governance and technical standards regarding high-autonomy agentic AI at the operating system level.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.