OpenAI dots stalls during its first public demo
IT Home reports that Dottie failed to respond to a launch-preparation request on stage. The cause of the interruption has not been established by the sources reviewed.
What happened on stage
OpenAI’s first public demonstration of dots stalled when the agent did not respond to a task, according to IT Home’s September 30 report, republished by Phoenix. ChatGPT product lead Holly Li asked a dot named Dottie to prepare the launch of a fictional music app. The report says the agent failed to reply and the demonstration paused, prompting a presenter to joke that Dot was slow that morning.
The report was published at 03:29 China time on September 30, corresponding to September 29 in UTC. This is the dots launch demonstration, not an older GPT-5 or Operator incident.
Why this matters for an always-on agent
Dots is pitched as a persistent agent with its own cloud computer that can continue work between conversations. A stalled live interaction raises a practical question: how clearly does it signal that a delegated task has stopped, and how easily can a user recover? That question belongs alongside the product’s launch and availability details.
What the evidence does not establish
The incident description here is attributed to IT Home; AITrending has not independently reproduced the failure. The reviewed sources do not establish whether the interruption came from the model, the connection, the demo setup or another component. One failed demonstration is not a measured failure rate for the released product. This story remains developing pending a supported explanation or follow-up reliability evidence.