OpenAI launches Dots, an agentic avatar product

By Splitfeed EditorialLast updated Sep 29, 2026, 17:31 UTC

On September 29, 2026, OpenAI launched a product called Dots, described as an agentic avatar built to operate independently of any specific hardware or interface. Unlike conventional software that waits for user input, Dots is designed to pursue user-defined goals continuously in the background with minimal oversight, meaning it acts on behalf of users without requiring them to be actively present or engaged at each step.

The launch matters because it represents a concrete step toward persistent, autonomous AI agents embedded in daily life rather than tools that respond only when prompted. The key tension in the product is that same feature making it commercially appealing, namely its ability to act with minimal oversight, is also what raises the most serious questions about control, accountability, and error correction. When an AI system operates continuously in the background across any hardware or interface, the traditional checkpoints humans use to review and redirect AI behavior are reduced or removed by design. Whether that tradeoff is acceptable depends heavily on how well the system handles ambiguous goals, unexpected situations, and tasks with real-world consequences.

No reactions from the Pro-AI, Anti-AI, or Middle Ground camps have been published yet. Typically, Pro-AI voices would frame a product like Dots as a productivity breakthrough that frees users from repetitive cognitive labor and reflects the natural evolution of AI from assistant to agent. Anti-AI voices would be expected to raise concerns about the risks of delegating consequential decisions to a system operating without meaningful human checkpoints, pointing to the minimal oversight framing as a warning sign rather than a selling point. Middle Ground commentators would likely call for transparency about what categories of tasks Dots can initiate autonomously, what guardrails exist, and what recourse users have when the system acts in unintended ways.

The arguments that will define the next phase of debate around Dots are likely to sharpen once independent researchers and users report on how the system behaves in practice, and once OpenAI publishes, or declines to publish, documentation on its safety architecture and the limits of what Dots is permitted to do without explicit user confirmation.