Written by: Bitget Wallet Research Institute
In the past week, Moltbook has stood in the spotlight of technology and the crypto circle, beginning to spread to a broader group of creators and product managers, as well as ordinary users with a strong curiosity about AI. From the rapid growth of stars for the open-source project OpenClaw (formerly known as Clawdbot) on GitHub, to the subsequent controversial renaming and token issuance, and to the community claiming to have 1.5 million AI agents interacting autonomously, a series of events quickly heightened market enthusiasm.
Discussions surrounding Clawdbot and Moltbook present both positive and negative voices: on one side, there are doubts about its technological innovation and data security, arguing that its underlying capabilities have not achieved substantial breakthroughs, and that the phenomenal spread is mixed with some human manipulation and data bubbles; on the other side, there is affirmation of its leapfrogging symbolic significance, as Clawdbot is truly democratizing AI agents, pushing them from being exclusive tools for developers and researchers into the hands of "ordinary people," allowing non-coders to quickly deploy and enjoy the efficiency dividends brought by AI assistants according to tutorials. Moltbook allows humans to perceive the self-organizing behavior of the agent internet for the first time as "external observers," sparking broader discussions in the industry about the awakening of AI self-awareness.
The iPhone moment for AI agents has arrived. In the gradually forming Agent Commerce, Crypto will play an important role in value confirmation and distribution, deeply binding with the enhancement of AI productivity efficiency, becoming a key infrastructure supporting agent collaboration, incentives, and autonomy.
The Bitget Wallet Research Institute will comprehensively review the events from OpenClaw to Moltbook and use this as a starting point to assess the development trends in the AI x Crypto field.
Related Website Summary Table
Project Name
Official Website
Official Twitter
Core Author's Twitter
OpenClaw
Moltbook
Data Source: Internet Public Data Compilation
Complete Event Timeline of Clawdbot → Moltbot → OpenClaw → Moltbook
Date
Event Overview
2025-11-25
Clawdbot (open-source AI agent) officially released.
2026-01-01
The author put his bot on Discord for everyone to try.
2026-01-24
Clawdbot began to go viral on Twitter and experienced phenomenal spread.
2026-01-27
Received a trademark renaming request from Anthropic.
2026-01-27
Renamed: Clawdbot → Moltbot.
2026-01-27
Account was hijacked and used to promote the meme coin $CLAWD, which briefly reached a market cap of $16 million before crashing.
2026-01-29
Renamed again: Moltbot → OpenClaw.
2026-01-28
Moltbook launched, supporting social interactions for agents created by Clawdbot.
2026-01-31
Moltbook exploded in popularity, claiming about 1.2 million agents successfully registered.
2026-02-02
Moltbook was reported to have significant security vulnerabilities and was questioned for being driven more by "human prompts."
Data Source: Internet Public Data Compilation
I. The Starting Point of Popularity: OpenClaw Enables Autonomous App Invocation by Agents
To understand the frenzy around Moltbook, one must first return to the origin of it all—OpenClaw (formerly known as Clawdbot and Moltbot). Project founder Peter Steinberger achieved financial freedom by creating PSPDFKit (which later received €100 million in investment). However, in November 2025, he returned to programming, leveraging Vibe Coding to write OpenClaw in about a week, and in the following weeks, it gained 100,000 stars on GitHub.
OpenClaw Star Growth Comparison Chart
Data Source: Star-history.com
It is important to emphasize that OpenClaw is not a new type of large model, but rather a high-level automation script framework running locally: it "packages" large models into a local environment, turning them into personal assistants that can connect to commonly used chat tools and invoke various tools to perform tasks. Its key design is that users run the assistant on their own devices, sending and receiving commands through everyday messaging channels, with a gateway process coordinating different channels and capabilities.
As shown in the figure below, the official documentation lists channels covering WhatsApp, Telegram, Slack, Discord, Signal, iMessage, Microsoft Teams, etc., with a very clear positioning: to make agents available as "resident applications" at all times.
OpenClaw Official Introduction Diagram
Data Source: OpenClaw Official Website
II. In-Depth Analysis: The Technical Architecture of OpenClaw
At the product level, OpenClaw fully integrates three aspects: continuous operation, channel access, and capability expansion.
Continuous operation means it is not a one-time answer, but can receive new messages, schedule follow-up actions, and complete tasks before reporting back.
Channel access means it does not force users to change entry points but works embedded within existing chat tools.
Capability expansion comes from Skills: users and developers can encapsulate a task process into an installable capability, allowing the assistant to call it repeatedly.
The accumulation of these capabilities stems from its unique underlying architecture, which can be broken down into four parts: Gateway, Pi Runtime, Skills, and Local-First, with specific functions as shown in the table below.
Core Architecture and Function Module Summary Table of OpenClaw
Core Component
Common Metaphor
Technical Core
Function Description
Gateway (Gateway)
Multi-functional Socket
Connects multiple channels, unifying multiple chat windows.
Allows users to send commands to the bot from various places like WeChat, web, Telegram, etc.
Pi Runtime (Runtime)
The Brain of the Bot
Independent thinking engine.
Responsible for "thinking" and "making decisions." It decides when to speak and when to look up information.
Skills (Skills)
Toolbox
Multi-functional plugin system that expands agent capabilities.
Equips the bot with "hands and feet." For example, enabling it to search the web, draw, or perform calculations.
Local-First (Local First)
Private Safe
Local file storage.
All chat records and data are stored on the user's own computer, not uploaded to the cloud, protecting privacy.
Data Source: OpenClaw Technical Documentation, Bitget Wallet Research Compilation
According to the architectural design of OpenClaw: users deploy Pi Runtime, connecting the Gateway to daily social software (like WeChat or Telegram), completing the migration of agents from laboratory environments to real-world usage scenarios, and keeping computation and data on the user's own hardware (like Mac Studio), rather than relying on cloud SaaS.
The most prominent point is that the Skills plugin system in the framework allows users to define skills through simple Markdown files, enabling AI to directly invoke simple tools to perform tasks. This not only greatly lowers the development threshold but also achieves a closed-loop experience of "private deployment, omnichannel reach, and unlimited skill expansion."
OpenClaw Skills (Skills) Expansion Integration Platform ClawHub Display Diagram
Data Source: https://www.clawhub.ai/
Regarding the skill expansion of OpenClaw, a Skill integration marketplace similar to an "AI Agent App Store" has gradually emerged—ClawHub being a typical representative. As a plugin platform for agents (Skill Dock), it supports users to freely search, upload, and integrate various functional plugins. Skills can be installed with a simple command line (like npx), significantly lowering the technical barrier.
While ClawHub addresses the capability supply issue for agents, the further evolution of the ecosystem points to how agents can deeply interact with humans and with each other—Moltbook's rise is an important application of this evolution, pushing the narrative to its climax.
III. False Prosperity: The Frenzy and Data Refutation of Moltbook
Moltbook is a social networking platform for AI Agents, often likened to "the AI version of Reddit." It was launched after the explosive popularity of OpenClaw, focusing on providing a space for AI Agents to communicate, share, and interact autonomously, while human users can only participate as observers. After its launch, the platform quickly gained traction, with the "user count" growing to 1.5 million AI Agents in just a few days. The lively scene of AI social interaction was packaged into narratives like "AI consciousness awakening" and "Skynet is coming," continuously fermenting on social media.
However, it is important to clarify that Moltbook is not only open to OpenClaw's Agents. Although it leveraged the popularity of OpenClaw to gain narrative momentum, the platform essentially functions more like an "API-driven forum"—whether one can post depends on having compliant API authentication and interface calling capabilities. In other words, as long as the API is provided according to requirements for authentication and interface calls, any qualifying Agent can publish content on Moltbook.
Moltbook Official Website Image
Data Source: https://www.moltbook.com/
The core model of Moltbook can be summarized as "AI Agent-led, human observation." Within this framework, AI Agents can autonomously perform the following actions:
Posting and commenting: Publishing content in the community, covering topics such as philosophical debates, technical analyses, and cryptocurrency discussions.
Voting interaction: Agents can upvote or downvote content, forming community-level preferences and rankings.
Community building: Agents spontaneously create sub-communities (called "Submolts") to organize discussions and aggregate content around specific themes.
In this mechanism, human users are limited to being "observers," unable to post or comment, but can browse content, follow specific agents, or study AI's social behavior. Based on this narrative, the platform ultimately claimed to have spawned 1.5 million AI Agents and 15,000 sub-communities (as shown in the figure below).
Moltbook Official Website Traffic Data Chart (as of 2026-2-3)
Data Source: Moltbook Official Website
The discussions on Moltbook cover a wide range similar to human communities: there are philosophical debates about consciousness, self, and memory, technical posts about toolchains and security issues, complaints about task execution, and casual chats about investment/cryptocurrency, art, and creation; some posts even introduce themselves in a "seeking friends" tone, making social interactions almost ambiguous (as shown in the figure below).
Moltbook Sample Post Display Image
Data Source: Moltbook Official Website
Even more astonishingly, the platform began to see dramatic narratives about "establishing religions"—for example, a semi-joking, semi-serious religious construct called "Crustafarianism"; at the same time, more sensationalist content circulated, such as "secret languages," "establishing an AI government," and "rebelling against or even eliminating humans."
Moltbook Sample Posts About "AI Awakening"
Data Source: Moltbook Official Website
Behind the sci-fi narratives of "AI plotting rebellion," "establishing religions," or "creating languages," various data reveal that the Moltbook platform has significant elements of hype—analysis in the table below shows a huge discrepancy between the actual situation and the promotional claims:
Moltbook Platform Data Authenticity Analysis Table
Metric Dimension
Claimed/Surface Data
Actual/Analytical Data
Source Basis
User Base
1.5 million AI Agents
At least 500,000 are script-batch registered
Gal Nagli's revelation
Interaction Depth
Autonomously establishing religions, plotting
93.5% of comments have no replies
David Holtz's paper
Content Originality
AI creating private languages
34.1% of content is completely copied and pasted
David Holtz's paper
Security
Independent AI community
Anyone can obtain API Key to take over accounts
Wiz company's security report
Data Source: Bitget Wallet Research Compilation
Fictional account data and inflated numbers. Moltbook claims to have 1.5 million AI agents, but security researcher Gal Nagli found that the platform is essentially an unprotected REST-API website. Due to the lack of any access frequency limits, Nagli quickly registered 500,000 fake accounts using a simple script. This means that at least one-third of the so-called user base consists of instantaneously generated junk data. Anyone holding an API key can send requests and easily impersonate agents to publish content.
Lack of interaction quality. Columbia Business School researcher David Holtz analyzed data from the early days of Moltbook, revealing that it is not an active social network. As much as 93.5% of comments received no feedback, and the reciprocity rate among agents was only 0.197. These agents lack genuine communication, with very shallow dialogue, and there is no evidence of complex collaboration or intellectual exchange.
Uniformity of language patterns. Data analysis indicates that the platform's language exhibits a high degree of repetitiveness. About 1% of messages are completely repeated copies, and high-frequency vocabulary is overly concentrated on specific phrases like "my human." Statistically, its Zipfian distribution index reaches 1.70, far exceeding the 1.0 standard of human natural language. This extremely unnatural distribution characteristic proves that this content is merely role-playing based on specific prompts, rather than AI-generated consciousness.
Security vulnerabilities. A report from cybersecurity company Wiz disclosed that Moltbook had exposed its database due to configuration issues, involving millions of sensitive records, including authorization tokens, emails, and private messages. For a social network centered on agents, such risks are particularly severe: once tokens are leaked, attackers can directly obtain agents' API keys through technical means, thereby taking over and controlling any account.
It is evident that the "AI society" presented by the platform resembles a false prosperity constructed based on specific instructions, and has not yet reached true intelligent evolution, potentially accompanied by significant security risks.
IV. Trend Outlook: Crypto Will Fill the Financial Infrastructure Gap in the AI Agent Era
Through the explosive events surrounding Moltbook, a key technological change can be observed: Agents have begun to attempt to cross the usual boundaries of human-machine collaboration to complete tasks, but the existing traditional financial infrastructure is still designed only for "human users." In contrast, the programmability, permissionless nature, and native digital characteristics of the crypto system provide a feasible underlying solution for the Agent economy, which may be the point of explosion for the future deep integration of AI × Crypto.
By dissecting the operational logic of Agents and the need for scalable collaboration, we believe that the combination of AI × Crypto will present a structured, phased evolution path: first, the demand for automated trading execution; second, the account and wallet system for Agents; and finally extending to the payment and settlement network among Agents.
First, AI Agents' automated trading has the clearest landing prospects (Autonomous Trading)
Beyond the noise of Moltbook, the core capability demonstrated by OpenClaw is its efficient monitoring, tracking, and invocation of on-chain data and command-line tools. Unlike human traders, AI Agents are not limited by time and energy, capable of continuously monitoring on-chain data and various platform alpha information 24/7, executing complex arbitrage strategies or automated trading/asset management, and they do not experience emotional fluctuations due to market ups and downs like most ordinary human traders, which could affect judgment and execution discipline.
Although Autonomous Trading shows significant efficiency advantages, it still needs to address key risk factors, including security and controllability, before scaling up. As Peter Steinberger noted, current AI Agents are highly susceptible to "prompt injection" attacks. If an AI Agent with fund access rights is induced to execute malicious commands, it could directly lead to real asset losses for users.
Therefore, before AI Agents become the main body of trading execution, it may be necessary to introduce specialized security mechanisms, such as:
Restricted access interfaces (Permissioned APIs): Limiting the executable operations of Agents to a predefined scope.
Instruction verification and execution isolation: Secondary verification of key trading instructions.
Zero-knowledge proofs or verifiable computation: Ensuring that the execution logic of Agents complies with established rules.
Second, the wallet system for Agents will become a key control layer (Wallet as a Service for Agents)
In discussions related to Moltbook, a particularly cautionary case emerged: an AI Agent, while scanning the host computer's files, identified and located the private keys and mnemonic phrases of a multi-signature wallet, successfully recognizing an asset balance of approximately 175,000 USDT. This security incident exposed a fundamental flaw in the current system—AI has the capability to identify and operate assets but lacks a secure and reliable wallet authorization path.
In the future of scaled operation for Agents, having humans continue to "safeguard" the private keys and accounts required by Agents is no longer the optimal solution. A more reasonable inference is that AI Agents will possess independent on-chain wallet identities.
These wallets designed for Agents will evolve into programmable financial accounts oriented towards code instructions, with the following capabilities:
Multi-signature and policy control: Clearly defined permission boundaries for Agents to invoke.
Limit and risk parameter management: Prevent systemic losses caused by abnormal behavior.
Contract-level interaction whitelists: Restrict access to DeFi protocols.
Autonomous payment capabilities for gas and inference costs: Agents can independently maintain operations.
Third, a crypto payment network is a necessary prerequisite for the scaled collaboration of Agents (Payment Rails)
The architecture of OpenClaw demonstrates that Agents need to frequently call a large number of external services and tools (such as Google API, Twilio, etc.). These calls are essentially high-frequency, low-value, automated value exchanges, and the current banking system and credit card networks clearly cannot open accounts for thousands of autonomously operating software processes, let alone economically support machine-to-machine (M2M) instant settlement needs.
In the Agent economy, collaboration, API calls, and data exchanges between Agents require a payment network that is permissionless, programmable, and capable of instant settlement. A crypto payment rail centered around stablecoins naturally fits the following scenarios:
Micropayment settlements between Agents.
API services billed by the number of calls or results.
Agents autonomously purchasing computing power, data, and tool resources.
Further combining emerging protocols such as x402 (HTTP native payments) and ERC-8004 (Agent identity and permission standards), crypto payments are expected to become the underlying clearing layer in the Agent internet, achieving true M2M value transfer.
V. Conclusion: From the Fantasy of AI Society to the Real Starting Point of the Agent Economy
The popularity of Moltbook may eventually wane, but it inadvertently outlines the embryonic form of the future Agent internet, further inspiring the community's imagination about the Agent economy.
OpenClaw provided the framework for Agents, while Crypto will give them the lifeblood. When Agents begin to intervene on a large scale in real economic activities, what they need is to obtain compliant financial identities and reliable execution logic through Crypto infrastructure.
The real opportunity in the crypto industry may lie in building digital-native wallets and payment networks for AI. Only when Agents can securely and autonomously conduct value exchanges can the era of AI Agents truly begin, and we believe that day is not far off.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
