1. Microsoft Chose OpenClaw to Power Scout, Its New Always-On Work Agent

At Microsoft Build on June 2, Microsoft unveiled Scout, a new "Autopilot" category AI agent that runs persistently inside Microsoft 365 — handling scheduling, email drafting, expense reports, calendar prep, and more across Teams, Outlook, and OneDrive, all without constant prompting. The kicker: it's built on OpenClaw. Rather than engineer an enterprise agent framework from scratch, Microsoft adopted OpenClaw as the foundation and contributed governance, sandboxing, and policy-conformance features back upstream to the project. Each Scout instance gets its own Entra identity, verifiable audit logs, and scoped data access controls. On X, the reaction ran from "this is the biggest validation yet" to wry jokes about Microsoft taming a "rogue data-leaking robot" — but even the skeptics admitted that a Fortune 500-targeted product shipping on your open-source framework is a different kind of endorsement than GitHub stars.

msdynamicsworld.com/story/microsoft-introduces-scout-openclaw-based-always-personal-ai-agent


2. Google Built "Gemini Spark" as Its Answer to OpenClaw — and the Community Took a Victory Lap

Announced at Google I/O on May 19, Gemini Spark is Google's cloud-based, always-on personal agent powered by Gemini 3.5 Flash — explicitly positioned as Google's response to the OpenClaw movement. It mirrors OpenClaw's core promise (24/7 background tasks, Gmail/Calendar/Drive integration, acts without a prompt) but runs on Google's cloud VMs instead of your own hardware. Wired ran the headline "Google's Response to OpenClaw's 24/7 AI Agent," and The Verge covered it alongside Microsoft Scout as part of a new "always-on agent" product category. On X, OpenClaw community members reacted with their signature mix of pride and pragmatism: two of the biggest tech companies in the world just announced products to do what the community has been running on Mac Minis for a year. Several posts noted OpenClaw still wins on privacy, model choice, and cost — but the imitation-as-flattery point landed loud.

wired.com/story/googles-response-to-openclaws-24-7-ai-agent


3. The Skill Workshop in v2026.6.1 Fixes the Thing Everyone Was Afraid Of

One of OpenClaw's longest-standing anxieties has been: what happens when an agent decides to silently rewrite its own skill? The new Skill Workshop, shipping in v2026.6.1 (stable, early June), answers it with a formal proposal/review/approval loop. Instead of directly overwriting SKILL.md, agents now file a proposal — users browse a review queue in the Control UI, make revisions, apply or reject, with full rollback metadata preserved and quarantine paths for anything suspicious. @iamlukethedev called it "a big milestone" and highlighted the adjacent reliability wins shipping alongside it: faster startup, steadier Telegram and WhatsApp delivery, and cleaner agent recovery from interrupted calls. The official @openclaw announcement thread drew strong engagement, with the comment thread converging on one theme: this isn't just a safety feature — it's what makes OpenClaw feel like a platform you can trust rather than a personal experiment you babysit.

x.com/openclaw/status/2062291555734696268


Spotted something great on X about OpenClaw? Share it — the community is the best part.