OpenAI Dots Privacy, Security and Safety

OpenAI's public materials for Dots emphasise user-controlled permissions, approval rules and separation of proactive (read-only) background activity from full agent actions.

Controls described

User selection of which plugins/apps a Dot may use
Rules that allow, require approval for, or block actions
Ability to inspect the Dot's cloud computer
Read-only restriction on unattended proactive research
Shared plugin permissions across ChatGPT surfaces where applicable
Memory interaction with ChatGPT memory settings

Age

Dots were described as not available to users under 18 at launch.

Data and training

Business, Enterprise and Edu workspaces follow OpenAI's standard non-training defaults for those products unless the customer changes settings. Personal plans are governed by the user's "Improve the model for everyone" preference for whether interactions may be used to improve models.

Granting a Dot access to email, calendars, code or business systems carries real operational risk. Permission hygiene and approval rules matter more than marketing language. Review OpenAI's Dots privacy/security help articles for the latest statements.