Microsoft confirmed to The Information that it is building new agentic features for Microsoft 365 Copilot inspired by OpenClaw, with security controls explicitly designed to address the open-source runtime’s risk profile. TechCrunch reported the story on April 13, noting the features “would be geared toward enterprise customers, with better security controls than the famously risky open source OpenClaw agent.”
This is a distinct development from the previously reported story about Microsoft integrating the actual OpenClaw framework into M365 Copilot. Microsoft is building its own agent with OpenClaw-like capabilities from scratch, not wrapping the existing open-source runtime in an enterprise layer.
The “Yet Another” Problem
TechCrunch’s headline captured the developer community’s frustration: “Microsoft is working on yet another OpenClaw-like agent.” The new effort joins at least two other agentic products Microsoft has shipped in 2026.
Copilot Cowork, announced in March, takes actions inside M365 apps using a “Work IQ” intelligence layer. It runs in the cloud and is powered partly by Anthropic’s Claude, which Microsoft added as an option after partnering with Anthropic late last year.
Copilot Tasks, released in preview in February, handles task completion for prosumers, from organizing email to managing travel. It also runs in the cloud.
The new proprietary agent would be the third agentic product, and the first designed to run locally rather than in the cloud, according to TechCrunch. The key differentiator from its siblings: “an agent that can complete multistep tasks over long periods,” effectively an always-on autonomous worker rather than a task-by-task assistant.
Security as the Selling Point
Microsoft’s framing positions the new agent as a direct response to OpenClaw’s security liabilities. The open-source project has faced nine CVEs in April 2026 alone, and a Qualys case study published April 14 demonstrated how a single OpenClaw vulnerability chains into full domain controller compromise.
A Microsoft spokesperson told TechCrunch: “Across our work, we are continuously experimenting as we bring broader orchestration and autonomy to our enterprise and consumer AI experiences while staying anchored in security, governance, and trust.”
The AI Insider reported on April 14 that Microsoft is “testing new AI agent capabilities within Microsoft 365 Copilot, aiming to introduce features similar to tools like OpenClaw while prioritising enterprise-grade security and control.”
The Mac Mini Factor
OpenClaw runs on Windows, but the Mac Mini has become the go-to hardware for OpenClaw users, driving unexpected sales of Apple’s small desktop, as TechCrunch noted. Beyond the security argument, Microsoft has a platform incentive: a proprietary local agent that runs best on Windows could pull users back from the Apple hardware that has benefited most from OpenClaw’s adoption.
Microsoft is expected to demonstrate the new agent, or an upgraded version of one of its existing agentic tools, at its Build conference in June, The Verge reported.
Validation Signal and Competitive Threat
Microsoft building its own OpenClaw-like product sends two signals simultaneously. For the OpenClaw project: the architecture is now the baseline for what enterprise agents should do. For companies building on OpenClaw, including Diana Intelligence Corp, which launched a managed OpenClaw SaaS today: Microsoft is coming for the same enterprise buyer with a proprietary alternative backed by the M365 ecosystem.