OpenAI has turned ChatGPT into a login button. According to OpenAI’s help centre, people can now use their ChatGPT account to create or access accounts on outside apps, the same job Google and Apple sign-in have done for years.
The button reads either “Sign in with ChatGPT” or “Continue with ChatGPT.” It shows up on participating partner sites and, where supported, when someone links a tool through the plugin directory inside ChatGPT. OpenAI’s page names six launch partners: Airtable, GitLab, HubSpot, Notion, Supabase, and Vercel. The feature already runs on OpenAI Academy and ChatGPT Sites, while OpenAI describes a gradual expansion to a chosen set of plugins and other partner sites, so many users will not see the button everywhere yet.
Two separate things are easy to blur here, and OpenAI keeps them apart. The first is identity. When someone signs in, the outside app receives only a name, an email address, and a profile picture if one exists. OpenAI’s page states that the sign-in alone does not share conversations, memory, files, tokens, or billing details.
The second is money. According to the same page, some partner tools let people run on their ChatGPT subscription’s rates, limits, and credits. After sign-in, a tool can ask permission to draw on the usage included in an eligible plan. Users can set limits on how much the tool consumes, and can finish the login without approving that request at all. OpenAI says this arrangement does not open up chats or memory either.
That second permission is the commercially interesting part. A developer who builds on top of ChatGPT sign-in could, if a user agrees, let that user’s existing plan pay for the AI features inside the developer’s product. Today most small software makers must either eat model costs themselves or pass them along through their own pricing. Letting the customer bring a subscription they already hold removes one of those choices. OpenAI has not published how partners are billed or compensated, and the help page does not say.
Companies get a say too. The help centre says the sign-in option is switched on by default for organizations that have not set an identity policy, while existing allow or deny lists stay in force. Only global admins can change tenant-level settings, in the External access section of the OpenAI Admin Console. For ChatGPT Business and Enterprise customers in an admin preview, delegated access for ChatGPT Sites and ChatGPT Ads starts off and needs explicit admin approval. Approval by an admin does not replace the individual member’s own consent.
The page also gives examples of extra permissions apps can request on top of sign-in. A ChatGPT Site can ask for access to a user’s connected apps, and HubSpot can ask for access to a ChatGPT Ads account. Each runs through its own authorization step.
There are limits to what can be concluded. The source is a help article, not an independent evaluation, and it carries no publication date, only a note that it was recently updated. The list of partners is short, and OpenAI describes the launch as limited, pointing other developers to an interest form. No adoption figures are given.
The strategic read is still clear. Google and Apple hold the default login on most of the web, and every sign-in gives them a view of which services people use. OpenAI is bidding for a slice of that role among people who already spend their working day inside ChatGPT, and it pairs it with a payment route the incumbents lack. Developers who sell AI features to ChatGPT subscribers should decide this quarter whether a “Continue with ChatGPT” button belongs next to their existing options, and whether they are comfortable with a customer’s usage limits being set by OpenAI’s plans rather than their own.
Reported from OpenAI’s help centre article on Sign in with ChatGPT, which carries no publication date.
