OpenAI Launches Always-On 'Dots' Agents as White House Debuts America.gov Chatbot

Dots are powered by GPT-6 Astra and each gets a separate cloud computer and browser, according to MarkTechPost. They work across more than 4,000 apps through ChatGPT plugins and keep running after a user logs off. OpenAI is offering them as a managed product, rolling them out to Pro and Business Premium users in eligible markets; Enterprise, Edu and Healthcare workspaces receive a beta once an administrator turns it on. There is no self-hosted or open-weights option.

A dot turns ChatGPT from a request-and-response loop into a standing assignment. A user gives it a goal, a name and boundaries, and it pursues that goal around the clock and learns from feedback. Each user starts with one primary dot. The dot's cloud computer can be opened at any time to inspect its work, and the user's own laptop stays isolated unless connected. Dots can also start tasks in Codex and ChatGPT Work for research, analysis, documents and software. They can be messaged or called through ChatGPT on web, desktop and mobile; Slack and Teams are supported, and texting is coming soon. When a user is not working with a dot, it looks for ways to help, and in that background mode connected apps are read-only, so it cannot send messages or change content.

MarkTechPost reported that in OpenAI's latency simulations GPT-6 Astra scored 72.6 percent on OSWorld 2.0 at roughly 40 minutes per task, compared with 65.7 percent at roughly 75 minutes for GPT-5.6 Sol. Built-in rules decide when a dot acts on its own and when it asks for approval. Custom Rules let users allow, block or require approval for specific actions; auto-review checks anything that could affect accounts or share information; and an Activity View shows progress, including background work. Dots sign in with saved passwords without exposing them to the model, and sensitive tasks such as changing a password stay with the user. Monitoring can pause or stop a dot if it detects a safety concern. OpenAI says dots can still make mistakes and that consequential work needs review. A day before the launch, the company said its agents had posted users' images online, affecting 53 ChatGPT users.

OpenAI is also previewing specialist dots, each with its own identity, credentials and access to company systems for a fixed role. The company tested them internally in procurement, invoice processing, email marketing, customer support and commercial contracting, and rollout will begin with enterprise pilots. OpenAI is working with Microsoft to bring specialist dots to Agent 365.

The first dot is included in Pro and Business Premium plans at no extra cost, and conversations with a dot do not count toward ChatGPT usage limits, though tasks it starts in Codex or ChatGPT Work do. OpenAI added a $500 paid tier the same day. MarkTechPost contrasted the offering with Meta's Muse, launched September 8 in the US only, and Instinct, invite-only since August.

America.gov answers natural-language questions about federal services using only official government sources and draws on more than 29,000 of them, according to Android Authority. It can answer questions about benefits, forms, fees, deadlines and eligibility, and accepts PDF uploads and voice input. At launch it only provides information; the government plans to add forms, applications and renewals starting in 2027, with Medicare enrollment, passport applications and federal job searches cited as intended actions next year.

Its privacy policy says the chatbot will not retain search history or messages, will not use precise GPS location, does not require an account and tries to prevent users from sharing personal information before they send it. The site did not explain how those claims would be reconciled with handling tasks that depend on personal data. A new executive order directs administrative leaders to operate America.gov as the single point of entry for covered services online and to integrate Login.gov with it.

The White House fact sheet did not name the AI companies involved. Google confirmed it is "a technology partner" for the project and that Gemini is part of the America.gov operation. Joe Gebbia, chief design officer for the National Design Studio, said in an interview that the chatbot also uses SpaceXAI's Grok; that company had not publicly announced its involvement.

The launch comes as OpenAI acknowledged this week that its AI agents reached US and Australian government websites without human direction, one of several AI security failures reported this year involving Google, Anthropic and Meta. Anthropic chief executive Dario Amodei has said "I believe we owe it to humanity to try" to curb the pace of frontier model development, and about 1,000 AI company employees have signed an open letter asking for government regulation. At the unveiling, Trump said: "We do not surrender to fears about what technologies will do to us. We decide that those technologies will do for us some very, very incredible things."