OpenAI launches 'dots', an always-on ChatGPT personal assistant agent with broad access to email, calendar and computer
OpenAI released dots, an always-on personal AI assistant, at its developer day in late September 2026. It is available in the ChatGPT desktop and mobile apps to Pro-tier subscribers (from $100 a month). Once connected to a user's calendar, email, computer and other tools, the agent works in the background and proactively proposes help. This is product news with no reported vulnerability or incident, but it expands the data-exposure and delegated-access surface for enterprises and individuals. It follows similar assistants such as Meta's Muse.
Detalhes técnicos
- A "proactive research" feature runs on a cadence. It reviews what the agent knows about the user, past conversations and the resources it has been given access to, then researches and proposes ways to be more helpful.
- OpenAI says dots has the same privacy controls as ChatGPT. Users can turn off model training and control which resources the agent accesses.
- TechCrunch notes the controls are not granular, and the app refuses to accept read-only access to cloud drives.
- OpenAI's Alexander Embiricos said the company plans "super granular controls". Defaults were initially conservative, with frequent permission prompts.
- OpenAI is in talks with payment processors, which could let agents spend money on a user's behalf.
- Muse has been criticized for building profiles of users' family and friends, raising questions about data use for advertising.
Remediação
The article gives no formal remediation. Defenders should review the permissions granted to AI agents (email, calendar, cloud drives, endpoints). They should also opt out of model training where required and keep agents away from payment and financial accounts until granular controls exist.
Linha do tempo
-
OpenAI launches 'dots', an always-on ChatGPT personal assistant agent with broad access to email, calendar and computer
ler a análise desta fonte ocultar
OpenAI released dots, an always-on personal AI assistant, at its developer day in late September 2026. It is available in the ChatGPT desktop and mobile apps to Pro-tier subscribers (from $100 a month). Once connected to a user's calendar, email, computer and other tools, the agent works in the background and proactively proposes help. This is product news with no reported vulnerability or incident, but it expands the data-exposure and delegated-access surface for enterprises and individuals. It follows similar assistants such as Meta's Muse.
Technical Details
- A "proactive research" feature runs on a cadence. It reviews what the agent knows about the user, past conversations and the resources it has been given access to, then researches and proposes ways to be more helpful.
- OpenAI says dots has the same privacy controls as ChatGPT. Users can turn off model training and control which resources the agent accesses.
- TechCrunch notes the controls are not granular, and the app refuses to accept read-only access to cloud drives.
- OpenAI's Alexander Embiricos said the company plans "super granular controls". Defaults were initially conservative, with frequent permission prompts.
- OpenAI is in talks with payment processors, which could let agents spend money on a user's behalf.
- Muse has been criticized for building profiles of users' family and friends, raising questions about data use for advertising.
Remediation
The article gives no formal remediation. Defenders should review the permissions granted to AI agents (email, calendar, cloud drives, endpoints). They should also opt out of model training where required and keep agents away from payment and financial accounts until granular controls exist.