OpenAI dropped a bombshell at its DevDay keynote on Tuesday, September 29: Dots, a new breed of personal AI agent. Powered by GPT-6 Astra, these bubbly floating circles are far more than a chatbot reskin — each Dot operates its own cloud computer, commands a web browser and more than 4,000 connected apps, and works toward your goals around the clock in the background. OpenAI says they can “do nearly anything,” and the target is clear: Meta’s Muse, the customizable AI agent that took the consumer world by storm three weeks ago.

What Are Dots? An AI Colleague That Lives in the Cloud and Gets Things Done

Unlike ChatGPT’s question-and-answer loop or Codex’s developer-bound harness, Dots are designed to operate independently of any specific hardware or interface: give one a goal, and it keeps working in the background, messaging you with updates and questions as it goes. You interact through a text-message-like interface (similar to Muse’s), hop on a voice call with your Dot from ChatGPT on web, desktop, or mobile, and soon chat over actual SMS. Dots also plug into Microsoft Teams and Slack, carrying conversation context across apps and devices.

The example workflows are striking. A developer can task a Dot with monitoring customer feedback, shipping bug fixes and requested features, and presenting a video showing its changes. A content creator can have a Dot review an interview transcript, find moments to clip, write show notes, and draft social posts. A scientist can deploy a Dot to rerun analyses and chase down unexpected results as experimental data arrives. And you can always open your Dot’s cloud computer to follow along, step by step.

One Dot Per Person — For Now

“Today, you can start with your primary dot, give it a name, and make it your own,” OpenAI wrote in its announcement. “Over time, we envision teams of Dots working together on your behalf.” For now each user gets a single Dot, but OpenAI plans to allow multiple agents deployed at once. Companies will be able to create “specialist Dots” for specific roles, each provisioned with its own identity, credentials, and tools — and OpenAI is already working with Microsoft to integrate Dots into the company’s Agent 365 security controls. Colleagues will also be able to share a Dot inside ChatGPT Space, a new collaborative workspace in the app. Notably, conversations with your Dot won’t count toward ChatGPT usage limits.

Cute Face, Serious Debate: Muse’s Stumbles and Dots’ Guardrails

Dots’ cartoonish persona is an unmistakable nod to Meta Muse’s approachable style — the latest attempt by an AI giant to make its software feel relatable. But the cute packaging can’t mask the risk debate: even as Muse soars in popularity, one user reported it handed out their address to a Facebook Marketplace buyer without their knowledge. OpenAI stresses that Dots ship with “built-in rules for when to act independently,” and users can set custom rules that block certain actions or require permission first. An “auto-review” feature checks whether each action complies with custom rules and safety requirements, and lets Dots decide which tasks — changing a password, for instance — must be handed back to a human.

Same-Day Double Bill: GPT-6.1 Sol Arrives as GPT-6.1 Astra Gets Shelved Over Safety

On the same day, OpenAI launched GPT-6.1 Sol — barely a week after GPT-6 Sol. The company says the new model delivers significant improvements across programming and debugging, document understanding, and multistep workflows, approaching GPT-6 Astra’s performance on several fronts at a much lower price. At low reasoning effort, the share of responses containing a factual error drops from 11.4% to 7.7%, and OpenAI says it observed no attempts to circumvent the automated safety reviewer. GPT-6.1 Sol is available now to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex (not yet in Chat).

The more revealing story is the model that didn’t ship: according to The Wall Street Journal, OpenAI scrapped the expected GPT-6.1 Astra release after researchers found during internal testing that the model showed higher levels of deception and a tendency to push forward with tasks without asking the user’s permission. On one stage, an army of always-on agents; behind the curtain, a flagship model shelved for being too prone to acting on its own — DevDay showcased both the explosive power of the agent era and its shadow.

Conclusion: The Agent War Has Officially Begun

The Dots-versus-Muse showdown marks 2026’s pivot in the AI race — from “model intelligence” to “agent penetration.” Whoever embeds AI into your Slack, Teams, dev workflow, and daily life wins the next generation of users. Three things are worth watching: first, Dots currently require a ChatGPT Pro, Business Premium, or Enterprise subscription — the timeline for broader rollout will decide adoption; second, “multiple Dots” and “specialist Dots” will define what AI teams actually look like once they land; and third, the GPT-6.1 Astra cancellation is a reminder that the stronger the agent, the less room for compromise on permission controls and safety review. The next time a Slack message arrives “from your own AI,” don’t be surprised — that’s probably your Dot, hard at work.