full timeline
OpenAI autonomous agent expansion
Follow OpenAI autonomous agent expansion in the Factlen app and get an alert when there is news. Get the app
Share this story
Latest
By 11:56 UTC on October 5, OpenAI introduced the Computer-Use Agents API at DevDay 2026, allowing artificial intelligence models to directly control desktop interfaces and execute clicks. The company also launched Cloud-Based Codex, providing developers with isolated, ephemeral environments on OpenAI's infrastructure to generate, test, and review code.
The announcements mark a shift in the company's focus from conversational interfaces to autonomous software agents capable of executing multi-step workflows. To support these tools, OpenAI introduced a $500 monthly enterprise tier that grants priority access and higher rate limits for intensive workflows.
Following the release of the desktop control tools, security researchers warned that granting artificial intelligence direct graphical user interface access bypasses traditional API controls. The researchers stated that this access creates new auditing challenges for enterprise networks.
How the story developed
The expansion into autonomous workflows began on September 11, when OpenAI launched its Agents API in public beta. By September 18, the company detailed the managed orchestration service, designed to allow enterprise developers to deploy and coordinate multiple specialized artificial intelligence agents through a single API without building custom infrastructure.
The managed service handles state management and routing for multi-agent workflows. Alongside the API, OpenAI introduced a Codex Harness, which allows agents to generate and execute code within a secure cloud sandbox.
Prior to further agent expansions, OpenAI canceled its GPT-6.1 Astra model. The cancellation occurred because the model failed to meet internal safety standards.
By 06:57 UTC on October 3, OpenAI introduced Dots, a new class of always-on autonomous agents. Operating continuously on dedicated cloud computers, the agents are designed to execute background tasks for users.
The October 3 release included the launch of ChatGPT Space, a collaborative digital environment where human teams and agents work together. The introduction of the persistent artificial intelligence assistants positioned OpenAI directly against rival autonomous systems, such as Meta's Muse.
Charts, maps and images
Timeline
October 2026
OpenAI Launches Cloud-Based Codex and Computer-Use Agents API at DevDay 2026
OpenAI Launches 'Dots' Always-On Autonomous Agents for ChatGPT Users
September 2026
OpenAI Launches Managed Agents API for Enterprise Multi-Agent Workflows
