On 10 November 2026, Salesforce disabled the Talent Rover connected apps and managed packages "as a precaution" after what it described as suspicious use of access tokens. The next day it did the same to the GoFundMe Pro connected app and its packages. Salesforce posted both on its Trust status site and said neither was a Salesforce platform vulnerability. Talent Rover was re-enabled the same day after the vendor's remediation. GoFundMe Pro was still off when Salesforce Ben reported it on 12 November.
If you used either product, you probably noticed: syncs stopped and users saw errors. If you didn't, this is still worth twenty minutes, because it is not a one-off. Third-party integrations are now the most common way into Salesforce data, and the pattern repeats: an attacker gets a vendor's OAuth tokens, uses them to pull data from that vendor's customers' orgs, and Salesforce cuts the app off to stop it.
This post explains what a disable means and gives you a short audit for your own org.
What "Salesforce disabled the connected app" actually means
A connected app (or External Client App) is how an outside service gets an OAuth token to your org. When Salesforce disables one platform-wide:
- Existing access and refresh tokens stop working for that app in every org, so the integration can't read or write data.
- Managed packages from the same vendor may be disabled too, so features inside Salesforce that depend on them can break.
- Your data and configuration are not deleted. When the app is re-enabled, the integration usually reconnects, though some vendors ask customers to re-authorise.
A disable is a containment step, not a finding that your org was accessed. Salesforce says it will notify affected customers directly if it finds unauthorised access to their data. Until then, the honest answer to "were we affected?" is: you need to look.
If you used an affected app: check these five things
- Watch Trust and your email. The Trust status message for the incident is the source for re-enablement. Salesforce notifies customers it identifies as affected.
- Ask the vendor for their incident notice. You need to know the window during which tokens may have been misused and what data the integration could reach.
- Find the integration user. Look at which Salesforce user the app's tokens belong to. The app could see everything that user could see. If it is an admin, or has View All Data, assume the reach was the whole org.
- Review that user's activity for the window. Login History shows API logins by app and IP address. If you have Event Monitoring (Shield or the add-on), the API and Report Export events show what was queried. Without it, Setup Audit Trail and field history are what you have.
- Revoke and reconnect cleanly when it comes back. Revoke the old tokens (Setup → Connected Apps OAuth Usage), reconnect with a fresh authorisation, and take the chance to cut the integration user's permissions to what the integration actually uses.
The 30-minute OAuth token audit (for everyone else)
You can't stop a vendor being breached. You can control how much a stolen token can reach and how many forgotten tokens are lying around. That is what this audit is for.
1. List every app that holds a token
Setup → Connected Apps OAuth Usage shows each app with active tokens and how many users hold them. Or query it:
SELECT AppName, User.Username, User.IsActive, LastUsedDate, UseCount
FROM OAuthToken
ORDER BY AppName, LastUsedDate
Go down the list and ask of each app: do we still use this, who owns it, and does it need the access it has?
2. Revoke tokens held by deactivated users
Deactivating a user does not always mean every integration they authorised has stopped. Any row above where User.IsActive = false is an access path that outlived the person. Revoke it.
3. Revoke tokens unused for 90 days
A token for an integration nobody has used in three months is pure risk: if the vendor is breached, it is a door into your org that no one will miss when it closes. Filter on LastUsedDate and revoke anything stale. If something breaks, the owner will re-authorise, and now you know who the owner is.
4. Shrink the integration users
Most integrations run as a dedicated user. Check each one:
- No admin profile. An integration user on System Administrator gives the vendor's token the whole org. Use the Salesforce Integration user licence and a minimal profile where you can.
- No View All Data or Modify All Data unless the integration genuinely needs every record. See who has Modify All Data for the query.
- Object access limited to what the integration syncs. A recruiting tool does not need Opportunities.
5. Tighten the connected app policies
For apps you control in Setup:
- Set Permitted Users to Admin approved users are pre-authorized and assign access through a permission set, so a random user cannot authorise the app with their own broader access.
- Use IP relaxation sensibly. If the vendor publishes static IP ranges, enforce them.
- Set a refresh token policy that expires tokens instead of valid until revoked where the integration allows it.
- Review OAuth scopes.
fullplusrefresh_tokenis the widest grant there is; most integrations needapiand little else.
6. Retire the old login methods
Integrations that still log in with a stored username and password are both the easiest to steal from and on a deadline: SOAP login() needs a new permission from 1 December 2026, and the OAuth username-password flow is retired on 20 February 2027. See our username-password flow guide for how to find and move them.
Make it a habit, not a one-off
This audit decays. New integrations get authorised, people leave, and vendors change. A quarterly review is a reasonable rhythm for most orgs; we covered how often to audit Salesforce security in more detail.
orgadmin.ai's Security Review runs the token part automatically. Its integrations checks flag:
- OAuth tokens held by inactive users (high severity)
- OAuth tokens unused for 90+ days (medium)
- Broad OAuth scopes on your org's own connected apps, assessed by AI
- Integration-style users with admin profiles (high)
- Insecure
http://remote sites and wildcard CORS entries
It is read-only. It shows you the finding and how to fix it in Setup, and the change stays yours to make. To act on inactive users in bulk, User Management shows who is still active, licensed and holding access.
The short version
When Salesforce disables a connected app, the vendor's tokens have probably been misused somewhere. Whether your data went with them depends on what that app's integration user could see. You can't control the vendor's security, but you can make sure each integration reaches only what it needs and that no forgotten tokens are waiting. Do that now, and the next "Salesforce disabled a connected app" headline is an inconvenience instead of an incident.