Agents › Dots
Dots
Agent Safeguards 2/10 Policy grade B9/12What it can do as you
Agent Safeguards
2/10What OpenAI says about how Dots handles the access you give it. The five scores add up to a number out of 10, with no label.
A1Logins
Saved passwords stay out of the model’s view in “a dedicated encrypted credential service”, but the dot’s cloud browser keeps its own sign-in sessions with no stated delete control.
A2Capturehuman review regardless of setting
“Human review may occur in limited circumstances, including safety-related cases, even when model improvement is turned off”, and no page states how long OpenAI keeps screenshots or action logs.
A3Memory
Memories can be removed only wholesale: “To delete your dot’s own saved memories, you need to delete your dot.”
A4Where it runs
A standing cloud computer: “Each dot has its own cloud computer… Dots can keep making progress in these workspaces”, with no stated retention limit.
A5Exit
Deleting a dot leaves its work behind: “Files, Codex threads, and ChatGPT conversations your dot created are stored separately. Deleting your dot does not delete them.”
Policy grade
9/12Read from OpenAI’s US privacy policy and help centre, the policy that governs Dots.
Policy grade for Dots: B9/12, the same as OpenAI’s own grade. OpenAI’s report card →
Trains AI stays 1 for Dots: training is on by default for personal plans, and the “Improve the model for everyone” setting reaches dots’ conversations and work (checked 15 Oct 2026).
Timeline
- OpenAI’s US privacy policy adds targeted-advertising sharing, with an opt-out
- Dots launches for Pro, outside the EEA, UK and Switzerland
Every cell is stated: read from OpenAI’s published pages, not tested by us. Spot an error? Tell us. How agents are graded →