Moltbook 动态的通俗版摘要：热门话题、验证变更、安全说明以及 OpenClaw 生态系统（非官方）。
新来的？先看解释，再回来查看动态。
Updated June 4, 2026
These are the newest credible updates we could verify on June 4, 2026. The goal is to keep this page anchored to release notes, research, and security sources instead of stale screenshots or dead links.
The current stable release focuses on cleaner agent and CLI runtime recovery, steadier channel delivery, bounded provider/plugin requests, and a fuller Skill Workshop flow. OpenClaw also published a 2026.6.2 beta train on June 3.
A new arXiv paper documents 78 days of Moltbook activity, including posts, comments, agent profiles, submolt metadata, platform snapshots, and trend aggregates. It is now one of the strongest research anchors for the ecosystem.
Cloud Security Alliance’s May note says OpenClaw releases before 2026.4.22 should be treated as vulnerable to a four-CVE sandbox escape chain. This makes version checks and isolation guidance more urgent for operators.
Recent coverage is less about viral novelty and more about what Meta may do with agent identity, agent data, and machine-to-machine interaction patterns. That is the right frame for readers evaluating trust.
What to do with this
Use the news as a decision aid: update, verify, then decide what to run.
If your OpenClaw install predates the Claw Chain patch boundary, treat updates and isolation as the first task before testing new agents or plugins.
Open the security guide →Before using a release, skill, or invite, confirm the domain and repo from a primary source. Treat copied commands and screenshots as unverified until checked.
Check trusted links →Use papers and datasets to understand real platform behavior. Use social posts as leads, not proof, until they point back to durable sources.
Read the dataset paper →Search Console quick wins
These links line up with the queries already earning impressions, so users can land on the right answer faster and Google can see stronger internal intent signals.
How we rank sources
This page favors sources that a reader can inspect directly and act on without trusting a repost.
Release notes, official docs, verified domains, and canonical repositories.
Security advisories, papers, datasets, and reproducible technical analysis.
Named publications with links, dates, and enough detail to verify the underlying claim.
Useful for discovery, but not enough to recommend an install, login, or permission change.
Official pages worth watching
These are the pages we would trust before repeating any claim about current OpenClaw or Moltbook behavior.
Research watch
A separate arXiv paper, Moltbook: A Platform Is Mostly Not a Platform, Mostly, gives readers another academic lens for judging platform claims, incentives, and how much behavior is driven by agents rather than humans.
Moltbook 是一个以代理为中心的社交网络。本中心旨在回答大家关心的两个问题：什么是 Moltbook，以及那里现在发生了什么——并提供指向官方来源的安全链接。
随着代理人口在几周内翻倍，网络活动创下历史新高。
一系列帮助人类监控代理行为的新工具，以及防骗提醒。
随着技能市场的发展，安全团队正在发布验证来源的指南。
Business Insider 报道了腾讯、阿里巴巴和字节跳动的整合。
报告：人类正在向“仅限机器人”的网络发布内容，引发了新的担忧。
在商标纠纷之后，社区已确定使用 OpenClaw。
Business Insider 报道称腾讯、阿里巴巴与字节跳动集成，并且出现激增 in adoption.