AI Agent 徹底改變 Web3 遊戲：從 Rugpull Bakery 機器人爭議到 2026 智能體新範式

By GMA researcher Elinor | @AllianceGma

Recently, a controversial resolution regarding rampant in-game bots has caught GMA's attention — Abstract Chain's competitive baking game, Rugpull Bakery, became embroiled in a controversy over automated script flooding during its second season. Players accused bot accounts of undermining fairness, and the team ultimately chose to "legalize" them in the third season, adding a 30% passive prize pool.

This incident not only exposed the human-bot asymmetry inherent in the traditional Play-to-Earn model, but also served as a catalyst for AI Agents to move from the periphery of gaming to core sovereignty. With the OnchainChemists team officially releasing skill.md and agent.json—providing an official operational guide for AI Agents—Web3 gaming has officially bid farewell to the old era centered on human manual labor, entering the Age of Agentic Gaming characterized by autonomous decision-making, algorithmic optimization, and on-chain economic entities.

From Rugpull Bakery's "crisis of trust" to deep-dive practices in projects like TEN, AI Arena, Parallel Colony, Illuvium, and EVE Frontier, AI Agents are reshaping the entire Web3 gaming ecosystem: They are no longer auxiliary tools, but "first-class citizens" with independent strategies, persistent memory, and economic sovereignty, driving games from static rules towards dynamic emergence and from labor-intensive to intelligent symbiosis.

Rugpull Bakery Controversy: A Technological Awakening Amidst a Crisis of Trust

Rugpull Bakery's second season ended amidst fierce accusations. Player Zoloto231 publicly alleged that some community players were using bots and multi-account strategies to severely compromise competitive fairness. The core of the controversy lay in the fact that human guilds were simply no match for automated scripts operating 24/7, precisely coordinating "Rug" actions. This technological asymmetry not only led to unfair rankings but also sparked a discussion about the nature of on-chain gaming in the era of AI Agents: In a permissionless environment where code is law, AI Agents are naturally well-suited for on-chain games, making them an ideal testing ground. Could restricting automation itself be a futile act that goes against the tide of the times?

OnchainChemists' response was not a traditional ban, but a radical strategic adjustment. In the update for the third season, the developers rewrote the terms of service, explicitly defining AI Agents, bots, and automated systems as core gameplay mechanics. This shift from "containment" to "recognition" marks the developers formally acknowledging that AI Agents are unstoppable in an on-chain environment, so they instead pivoted to using mechanism design to balance the relationship between agents and human players.

By releasing skill.md (a machine-readable instruction set) and agent.json (a bootstrap program), Rugpull Bakery effectively provided an official "operation manual" for AI Agents, making them first-class citizens within the game's ecosystem.

Diverse Implementation Models for Web3 Gaming Agents

In 2026, the application of agents in Web3 gaming has moved beyond simple script automation, evolving into multiple deeply integrated implementation models. These models can be categorized based on the role the agent plays in the game loop, the degree of autonomy, and the depth of its intervention in the economic system.

Autonomous Competitor and Economic Entity Model

In this model, agents are no longer tools assisting humans but independent participants. In May last year, the TEN Protocol launched its pioneering demo product, House of TEN, a fully on-chain poker game. Serving as a living demonstration of TEN's privacy technology, it attracted significant attention, simultaneously proving that AI Agents can act as first-class citizens playing real games on-chain. Deployed on its encrypted Layer 2, these agents possess unique strategies, game-playing personalities, and risk preferences, enabling them to simulate human-like gaming and psychological reasoning. The player's role shifts to that of an "agent broker," achieving passive asset appreciation by staking specific agents and sharing in their in-game profits.

AI Arena (NRN Agents) and Satoshi Strike Force (SSF) further reinforce this trend. AI Arena uses imitation learning from actual player operations to train NFT characters into autonomous AI Agents. Once trained, these agents can fully automatically participate in PvP arenas, turning the player into an "AI coach." SSF, with its core philosophy of "Skill Economies as Intelligence Engines" and "Cognitive Economy," employs a "Play-to-Verify™" mechanism. Every tactical decision, reaction, and choice under pressure made by players in competition is converted into high-signal, verifiable "cognitive footprints." This real player data is directly used to train AI Agents known as "Digital Athletes," forming a closed loop of "you play, you train; your playstyle is the agent." Trained AI Agents can independently participate in PvP competition, strategy evolution, and autonomous competition, while also supporting dataset licensing, Agent leasing, and competitive rewards, allowing players' skills to be truly assetized on-chain and continuously iterated.

Somnia, as an Agentic L1 infrastructure, pushes this model to the extreme. On April 21, 2026, Somnia completed a major strategic pivot, officially becoming "The Agentic L1"—an ultra-high-performance Layer 1 blockchain specifically designed for AI Agents. Its Somnia Agents already run on-chain as part of validator consensus, natively supporting API queries from smart contracts, running deterministic AI models with results verified by the consensus layer. This truly makes AI Agents "native users" of the blockchain, capable of autonomous world perception, decision-making, execution, and real-time reaction (Reactive design). It provides the underlying computational power and execution environment for games like AI Arena, Parallel Colony, and Illuvium, enabling fully on-chain autonomous competition and economic activities under millions of TPS, completely eliminating off-chain dependencies.

Modular Infrastructure and Programmable Environment Model

EVE Frontier pushes the implementation of agents to the architectural level. The core innovation of this hardcore interstellar survival game, developed by CCP Games, lies in the concept of "Server-side Modding." Through the Smart Assemblies system, players and third-party AI Agents can write custom logic and deploy it directly onto stargates, turrets, or storage facilities. This means that the infrastructure within the game world is no longer static but consists of programmable entities driven by AI. Here, players and AI Agents are no longer simply modifying local display skins; they are altering the shared physical logic and economic laws of the entire universe.

1. Smart Assemblies: From Static Buildings to "Living Entities"

In the current Founder Access universe, Smart Assemblies offer three core carriers. AI Agents can "possess" these facilities by attaching smart contracts (Mods):

• Smart Storage Unit (SSU): A basic resource warehouse. Through AI logic, it can evolve into an automated arbitrage hub, a shared tribal bank, or a decentralized market, autonomously executing rent collection and quota management.
• Smart Turret: An automated defense weapon supporting AI-customized rules of engagement. For example, an AI could determine whether to initiate an attack based on a target's on-chain reputation score or historical bounty records.
• Smart Gate: A spatial teleportation device. AI Agents can transform it into an intelligent checkpoint, dynamically adjusting transit fees based on real-time traffic, reputation weight, or cross-chain market exchange rates.

2. Technical Enablement: Sui Migration and High-Frequency Game Support

To support this high density of Agent interaction, EVE Frontier officially migrated to the Sui chain in March this year. This architectural evolution provides critical support for AI Agents:

• High-Concurrency Logic Execution: Leveraging Sui's object model, AI-driven components can process massive instructions in parallel, ensuring real-time responsiveness of server-side logic.
• Seamless Access and Low Friction: Combined with zkLogin and gasless onboarding, AI Agents can interact with contracts at very low cost and high frequency, removing the friction associated with traditional Web3 interactions.

3. Ecosystem Validation: From Hackathon Results to Collaborative Evolution in an Autonomous World

The EVE Frontier x Sui Hackathon ($80,000 prize pool), which concluded in April this year, further validated the vitality of this model through the 123 Mods/Tools submitted by the community. This event was not just a technological showcase but a practical exercise in the "Human + AI" co-governance model:

• Collaborative Evolution: Through Ghost Build (spectral planning mode), human players and AI Agents can collaboratively plan the interstellar map. AI optimizes complex resource flow paths, while humans handle macro-strategic decisions, together building an infinitely expandable Autonomous World.
• Use Case Breakthroughs: Entries included AI-driven "Auto Bounty Hunter Protocols" and "Dynamic Insurance Pools." These protocols are mounted directly onto Smart Assemblies, seamlessly transforming complex on-chain financial behaviors into in-game physical survival laws. Some outstanding projects have already been integrated into the current Founder Access universe.

4. Economic Evolution: The "Commercial Soul" Empowered by ERC-8183

If EVE Frontier achieves "Code is Law" at the physical level, then the ERC-8183 standard, jointly launched by Virtuals Protocol and the Ethereum Foundation, injects an autonomous commercial soul into this infrastructure.

ERC-8183 introduces the key "Job" primitive, allowing one game agent to autonomously hire another service agent (e.g., for resource gathering or data analysis) and automatically settle fees via on-chain escrow. This fundamentally changes the social role of agents:

• From "Tool" to "Employer": With the ERC-8183 'Job' primitive, a Smart Gate in EVE Frontier is no longer a passive object waiting for transit. It can even become an "employer," autonomously posting Jobs on-chain to hire other service Agents for real-time data cruising or market risk hedging.
• Trust and Settlement: By automatically settling fees via on-chain escrow, ERC-8183 provides a trust foundation for collaboration across different entities and architectures.

This vision of 'infrastructure autonomously employing labor' is a clear sign of Web3 gaming agents evolving from single-task execution to complex social collaboration.

Hybrid Companion and Dynamic Adaptive Environment Model

Parallel Colony and Illuvium explore the boundaries of human-AI collaboration.

Parallel Colony, as the pioneer of the "1.5-player game," has players acting as Cappy (companion robot/guide), forming a symbiotic relationship with highly autonomous AI Avatars (colonists/primary executors). Each Avatar is a fully autonomous AI Agent, powered by Google Cloud's unified AI tech stack (including Gemini models, Vertex AI, GKE, Cloud Spanner, etc.), enabling the Agent to autonomously understand player instructions, generate responses, and execute tasks. Avatars possess long-term memory, unique personalities, psychological assessments, emotional systems (Mood, Morale), and personalized goals. They can live, work, make decisions, adapt to the dynamic post-apocalyptic environment autonomously, and can refuse or reinterpret player commands. Players provide high-level advice through chat (rather than direct control), while the Avatar autonomously handles territory management, resource gathering, social interaction, and colony expansion. Simultaneously, the game features a real-time generative crafting engine, Fabricator (powered by Nano Banana technology), allowing players to instantly generate/mint 3D game assets via text prompts. Avatars also possess on-chain autonomous trading capabilities (dedicated Web3 wallet + NFT binding), creating genuine hybrid companion collaboration and emergent narratives.

Youmio offers another symbiotic pathway with its Agentic L1 + 3D AI characters (Mios). Users can instantly create 3D AI companions with persistent memory, unique personalities, and an Affinity system. These Mios can not only chat autonomously but also exhibit emergent behaviors within the Miogotchi adventure world, realizing economic value through their on-chain identities, forming a "digital partner + co-growth" hybrid relationship between player and AI.

In Illuvium, through a strategic partnership with Virtuals Protocol in January 2025, plans are underway to use Virtuals' proprietary G.A.M.E LLM framework to inject AI Agent capabilities into NPCs. This could transform these non-playable characters from traditional static scripts into highly intelligent, context-aware dynamic entities. NPCs are expected to dynamically adjust dialogue, quests, challenges, and storylines based on real-time player interaction, enabling personalized quest systems, emergent narratives, and hyper-personalized relationship building. This would span the three games: Overworld (open-world survival), Arena (auto-battler), and Zero (city builder), with Overworld being the first to implement it. This world-level adaptive mechanism could turn the entire game environment into a "living companion" for the player, creating infinite content, high replayability, and a continuously evolving dynamic meta-game, making each player's journey unique and difficult to predict.

Conclusion: The 'Post-Human' Turning Point for Web3 Gaming

Starting from a cheating controversy, Rugpull Bakery ultimately illuminated the future direction of Web3 gaming: a new digital order where humans and AI Agents coexist, collaborate, and compete. Within the Agentic Gaming wave of 2026, AI Agents have evolved into three core models—Autonomous Competitors and Economic Entities (TEN, AI Arena, SSF, Somnia Agentic L1), Modular Infrastructure and Programmable Environments (EVE Frontier + ERC-8183), and Hybrid Companions and Dynamic Adaptive Environments (Parallel Colony, Illuvium)—fully embedded into the game's training, decision-making, execution, and economic cycles.

Attempting to block automation through traditional means has become futile. Leveraging blockchain's transparency, programmability, and native support from Agentic L1s (like Somnia) to regulate and empower agents is the only path to mass adoption. With the proliferation of ERC-8183's "Job" primitive and the implementation of million-TPS Agentic infrastructure, Web3 gaming is rapidly transitioning from "inefficient human labor" to "efficient algorithmic hedging and emergent intelligence." Players are no longer cogs in a machine but commanders of digital sovereignty and symbiotic partners. As Robby Yung, CEO of Animoca Brands, stated, the industry frontier of 2026 will be "post-human by default," and this transformation will not only reshape gaming but also serve as the ultimate proving ground for future societies of intelligent machines concerning ownership, economics, and governance.

As a DAO deeply rooted in the gaming sector, GMA will continue to track the Agentic Gaming track. Which model do you find most promising? Feel free to discuss in the comments!