started · updated
OpenAI introduces Dots autonomous AI agents
OpenAI has announced Dots, a new class of AI agents powered by GPT-6 Astra designed to perform tasks autonomously between user conversations. Unlike standard chatbots that respond only to direct prompts, Dots operate on their own cloud computers and can connect to over 4,000 applications to pursue standing goals.
These agents are intended to act more like coworkers than reactive tools, capable of handling workflows such as drafting invoices from email threads or monitoring customer feedback to propose software fixes. For organizations, OpenAI is previewing specialist Dots with distinct identities and credentials, and is collaborating with Microsoft to integrate these agents with Agent 365 governance and security controls.
The technology introduces new considerations for privacy and supervision. While users can disconnect applications, information already incorporated into a Dot’s context remains. Additionally, OpenAI has established boundaries between authorized background tasks and unsolicited proactive research, the latter of which is restricted to reading permitted sources and cannot directly modify application content or control hardware.