OpenAI's Dots agents can read email 2FA codes with user permission, report says
In brief
- OpenAI launched its Dots AI agents at DevDay on September 29, 2026.
- Dots agents can read email verification codes and complete 2FA, but only with user permission.
- Safeguards include Auto-review, user-defined Custom Rules and limits on high-risk actions.
- Dots access is limited to ChatGPT Pro, Business Premium and an Enterprise beta.
What Dots are
According to Crypto Briefing's report, the agents run on OpenAI's GPT-6 Astra model. They're built to keep working in the background instead of only answering when someone types a prompt. Each Dot operates on its own isolated cloud computer, works independently and remembers context from one interaction to the next (a big part of the pitch for multi-step jobs).
Dots connect to over 4,000 applications through plugins, the report said, and they plug into work platforms people already use, including ChatGPT, Slack and Microsoft Teams.
The 2FA piece
The email-code feature goes after a familiar friction point. When a sign-up or login flow sends a code to a user's inbox, a Dot can grab it and enter it, provided the user has given permission. Crypto Briefing described the goal as workflow efficiency: multi-step tasks that would normally stall at a "check your inbox for a code" step can keep moving without a human stepping in.
That's a lot of access to hand an agent.
OpenAI paired the feature with several safeguards, per the report. There's an Auto-review process, Custom Rules that let users define which tasks are authorized, and restrictions on high-risk actions.
Those high-risk categories include transaction confirmations and changes to sensitive data.
Initial user feedback, as reported by Crypto Briefing, indicated that Dots hesitated to enter 2FA codes even when they had email access. The outlet attributed that behavior to built-in security configurations.
Who can use it
Availability is limited to ChatGPT Pro and Business Premium subscribers, with one Dot included in qualifying subscriptions, the report said. An Enterprise beta is also available.
Trial participants so far have mostly been Pro and Business Premium users. They've been testing how the agents handle multi-step work under the safety rules, and whether a Dot touches anyone's inbox at all still comes down to the permissions that user grants.
Frequently asked questions
How do OpenAI's Dots agents handle two-factor authentication codes?
According to Crypto Briefing, a Dot can read a verification code sent to a user's email during sign-up or login and enter it. It only does this if the user has granted permission. Initial user feedback reported by the outlet indicated Dots sometimes hesitated to enter the codes, which Crypto Briefing attributed to built-in security configurations.
What safeguards did OpenAI attach to the Dots email-code feature?
Crypto Briefing reported that OpenAI paired the feature with an Auto-review process, user-defined Custom Rules for authorized tasks and restrictions on high-risk actions. The high-risk categories include transaction confirmations and changes to sensitive data.
Who can use OpenAI's Dots agents?
Per Crypto Briefing, availability is limited to ChatGPT Pro and Business Premium subscribers, with one Dot included in qualifying subscriptions. An Enterprise beta is also available.
