AI News Feed
Market watch
Products & Applications

OpenClaw 2.0 launches with collaborative cloud sessions, stronger security

OpenClaw 2.0, the largest update to the open-source personal AI agent, adds collaborative cloud sessions and a rebuilt dashboard.

Unlike traditional chatbots such as ChatGPT, OpenClaw acts like a digital secretary that executes tasks on a user's behalf, living on a computer or in the cloud. The free, open-source AI agent is widely used as a coding agent or personal assistant to organize inboxes, manage calendars and check daily activity.

The OpenClaw team is branding the release as 2.0 even though the version number is 2026.8.1, because it includes major quality-of-life updates, better persistence, easier installation and increased security. The installation process has been streamlined so new and veteran users can detect existing ChatGPT or Claude subscriptions, API credentials and locally hosted models. Certain configuration options now default automatically, letting users start interacting with the agent sooner.

The dashboard has been rebuilt as a control center with chat front and center, giving users a familiar AI interface. They can start new chats, rejoin ongoing work and watch the agent working live. New interactive widgets can be brought into chats, pinned to dashboards and assigned actions triggered by events.

The release removes a previous limitation that kept groups from working with the same agent. Shared cloud sessions now allow team members to join ongoing sessions, continue where it left off, assign tasks with full context, and collaborate in real time on content and automations. According to the team, they use this "multiplayer" collaborative feature to build the OpenClaw framework code itself.

Search has been updated so users can find past conversations using exact text, words or phrases and reopen surrounding messages. Security improvements include explicit trust boundaries, least-privilege controls and stronger protection against prompt abuse. The system now controls access based on roles, tool handling, and rules for who can trigger it, what it can access, where code executes and where credentials are sent. Early versions of OpenClaw suffered from major security flaws, and the team says these updates reduce the attack surface, though users are still expected to harden installations and maintain strong authentication.

This is the first major version update since OpenAI Group PBC hired Peter Steinberger, the creator of the AI agent framework.