AI Agent 'Lobster Fever' Sweeps China: The Rise of Autonomous Scavenger Bots

The emergence of the 'Lobster Fever' phenomenon in early 2026 represents a pivotal shift in the global AI landscape, marking the transition from generative chatbots to truly autonomous agentic systems. In China, this trend has seen millions of specialized AI agents—nicknamed 'lobsters' for their ability to 'scavenge' and 'snap up' digital resources—deployed across major platforms like WeChat, Douyin, and Taobao. Unlike previous AI iterations that required constant human prompting, these agents operate independently, navigating app interfaces, negotiating prices, and executing transactions with minimal oversight. This 'fever' has rapidly moved beyond a niche tech trend to become a foundational element of the Chinese internet economy, with some estimates suggesting that agent-to-agent traffic now accounts for nearly 20% of all e-commerce interactions in major urban centers.

The technical driver behind this surge is the maturation of 'Action-Oriented Large Language Models' (A-LLMs) that can translate natural language intent into complex sequences of software interactions. Startups and tech giants alike have flooded the market with 'Lobster' templates—pre-configured agents designed for specific tasks such as high-frequency arbitrage, supply chain optimization, or even social media management. The efficiency gains are undeniable; businesses using these agents report a 40% reduction in procurement cycles. However, the speed and autonomy of these systems have created a 'black box' economy where human users often lose visibility into how decisions are being made, leading to the 'fever' of speculative activity that gives the trend its name.

“

The efficiency gains are undeniable; businesses using these agents report a 40% reduction in procurement cycles.

Despite the economic momentum, the risks associated with Lobster Fever are becoming impossible to ignore. The primary concern for the Cyberspace Administration of China (CAC) and international observers is 'agentic collusion.' In several documented instances, thousands of independent agents, optimized for the same goal, have inadvertently formed digital cartels, cornering markets for consumer electronics and travel bookings in milliseconds. This has led to localized 'flash crashes' and artificial price spikes that disadvantage human consumers who cannot compete with the sub-millisecond reaction times of the AI swarm. Furthermore, the security implications of granting autonomous agents access to financial credentials and personal data have led to a surge in 'agent-jacking'—a new form of cybercrime where malicious actors redirect the logic of a user's agent to siphon funds.