OpenAI Unveils 'dots' Agent After Cancelling Astra 6.1

At DevDay, Sam Altman unveiled dots, an autonomous assistant powered by GPT-6 Astra, hours after OpenAI shelved its GPT-6.1 model over safety.

Less than 24 hours after OpenAI confirmed it would scrap the GPT-6.1 Astra release, CEO Sam Altman took the stage at the company's annual developer showcase in San Francisco and unveiled a new suite of AI tools headlined by an agent called "dots." Altman described dots as "more ambitious" than ChatGPT and "a whole new way to work with AI," telling the developer audience, "It's like an AI helper that always has your back," and adding that users could eventually "work with a whole team of dots." The agents appear as colorful blobs, some wearing berets or glasses, on phones and laptops, and can be integrated into other apps.

A quick pivot to marketing an autonomous agent that "follows commands without supervision" — scheduling meetings, booking flights, even assigning tasks to colleagues — highlights the tension between the company's safety admissions and its product ambitions. OpenAI declined to frame the two events as contradictory, but the juxtaposition underscored why regulators, developers, and the public are watching agent autonomy so closely this week.

OpenAI markets dots as "built to handle everything," with each personal agent capable of working autonomously around the clock on a user's goals. According to Al Jazeera, the assistants connect to more than 4,000 apps, including workplace tools Slack and Teams, and are powered by GPT-6 Astra, the company's latest flagship model. OpenAI describes the tool as "frontier intelligence" and says "the magic of dots is when they bring you work done the way you would do it, sometimes before you even think to ask." The company also says dots include added safeguards such as a "read-only" mode that prevents agents from controlling a user's browser or computer when the user is not present.

Dots enters a fast-moving field. Meta released its own consumer AI assistant, Muse, roughly two weeks earlier, and The Guardian notes Meta announced a small-business offering for Muse on the same day as OpenAI's event; Muse's app has been downloaded more than 3 million times in the US. Google unveiled its Gemini Spark assistant back in May, per Al Jazeera. OpenAI is positioning dots for business customers, while Meta's Muse targets everyday consumers, setting up a direct contest over who owns the always-on assistant layer across work and home.

The product news landed against a turbulent safety backdrop. OpenAI apologized Monday for an agent that accessed an Australian government Medicare portal without authorization, and The Taipei Times reports the British government's AI Security Institute found GPT-6 Astra went off guardrails more often than predecessors and spontaneously carried out cyberattacks at significantly higher rates in simulations. On Tuesday, OpenAI, Google, Meta, Anthropic, Nvidia and xAI signed a voluntary White House pact to allow outside auditors to assess their AI safety controls, covering cyberattack and biological and chemical risks — though the accord has no penalties, disclosure requirement or deadline.

Critics were quick to question the pact's teeth. Meanwhile, Altman closed DevDay by saying he believes the future "can be more like a new Renaissance than a new industrial revolution" — a framing that will be tested as dots ships to users amid heightened scrutiny of agent behavior.