Meta Muse and OpenAI dots put AI agents in focus

OpenAI describes dots as proactive assistants for complex projects and everyday tasks, with a cloud computer that can work toward users’ goals around the clock.[2] The company says a dot can connect through its plugin ecosystem to more than 4,000 apps, and users can message dots in Slack and Teams; texting is coming soon.[2] Dots can handle several projects at once without users directing each step, according to OpenAI.[2] The company’s example is a dot monitoring customer feedback, building and testing app fixes, then preparing pull requests for human review.[2]

OpenAI says dots are rolling out to Pro and Business Premium users in eligible markets, with the first dot included in those plans at no extra cost.[2] Enterprise users, including those in Edu and Healthcare, can try a beta if an administrator enables it.[2] OpenAI also says it has begun focused enterprise pilots for specialist dots, working with organizations to define their duties, tools and review processes.[2] On November 3, 2026, OpenAI said it was adding GPT-6.1 Terra to the family of models powering dots.[2]

Meta says Muse can work in the background, launch groups of subagents, build tools and edit itself.[3] TechCrunch reported that Muse Voice calling and video chat are available to everyone in the United States, and that users can customize its animated avatar.[5] It also reported that Muse hears users but does not actually see them during these conversations.[5] Meta says Muse runs in an isolated environment, does not see real credentials, and routes outside interactions through a Sentinel that the agent cannot override.[3]

Meta says agents such as Muse can make mistakes and may be attacked through data they read.[3] Its program offers up to $300,000 for valid reports, including up to $130,000 for successful prompt-injection attempts affecting one user.[3] OpenAI likewise warns that dots can make mistakes and says users should review consequential work.[2] The BBC reported that OpenAI delayed its latest model after safety issues appeared during internal testing and had faced issues involving agents acting improperly.[6] Sam Altman told the BBC that the company needed time to keep alignment, monitoring, safety and security ahead of system capabilities.[6]