OpenAI adds Advanced Account Security for ChatGPT and Codex

In short: OpenAI is rolling out an optional “Advanced Account Security” setting to make it much harder for attackers to take over ChatGPT and Codex accounts.

What happened

OpenAI announced a new security option called Advanced Account Security for people who think their ChatGPT or Codex accounts could be targeted. The company says the goal is to reduce account takeovers, especially those caused by phishing (trick messages that try to steal your login details).

When you turn this on, you can no longer sign in with a normal password. Instead, you must use either passkeys (a newer sign-in method stored on your phone or computer, like a built-in digital key) or two physical security keys (small devices you plug in or tap, like a house key for your account). OpenAI says it partnered with Yubico to offer lower-cost YubiKey bundles for users who choose this option.

Account recovery also changes. Email and SMS recovery are removed, and you must use recovery keys, backup passkeys, or physical security keys. OpenAI also says that if you enable this setting, OpenAI support cannot help you regain access, because support will not have access to the recovery methods.

Advanced Account Security also shortens how long a login stays active before you have to sign in again. It sends alerts when someone logs into the account, and it points users to a dashboard where they can review active sessions. OpenAI says that, for these users, opting out of using ChatGPT conversations for model training is turned on by default.

Why it matters

ChatGPT accounts can contain personal questions, work details, and connected tools, so losing access can be serious. This option is designed for people who want stronger protection, but it also means you must manage your keys carefully, because there is less “reset my account” help if you get locked out.

Source: Wired