Key Takeaways:
- Mobile apps can be directly abused to aid in phishing attacks
- MDM and MTD don’t detect undisclosed mobile app vulnerabilities
- App store approval doesn’t equal enterprise security
- Mobile app vetting fills a critical gap for detection and prevention of sophisticated attacks

What the app is
It’s an SMS inbox app on Google Play, in the Communication category with tens of thousands of reviews and 10 Million+ Downloads. The publisher is one of the biggest names in computing, a company whose software is already on nearly every managed laptop and desktop. That pedigree is part of why an app like this gets installed, and approved, without a second look.
It sorts incoming texts by type, sets reminders for bills and appointments, and backs messages up to the user’s cloud drive. The store listing asks people to make it their default SMS app, and plenty do. That makes it the app holding their bank alerts, delivery codes and the one-time passcodes their employer’s MFA sends by text.
The app has been anonymized. Reach out to us directly to get the name and the de-anonymized risk information.
What we observed
One of the app’s features is a callback reminder. Miss a call and it can nudge you later: ” [CALLER] called you an hour ago”, with a CALL NOW button.
Q-scout’s manifest analysis flagged the component behind that reminder as exported with no permission attached. Exported means any other app on the phone can send it a message. No permission means the sending app needs nothing to do so. No install-time permission, no runtime prompt, nothing the user would see.
We confirmed it on a device. A second app sent the component one message carrying a name and a phone number, and the SMS app posted the notification below. The name and the number are both whatever the attacker chooses.
In the example below we used “Dad” and “Mom” with a generic corporate call-center number to showcase the risk. A real attacker would craft an easy-to-believe combo. “HR” and a local phone number from the office, or everyone’s favorite “your CEO’s name” and a request to buy gift cards for the next company party!
Tapping CALL NOW opens the phone’s dialer with the attacker’s number already filled in. The sender can also hide the “Remind me later” option, so CALL NOW is the only thing to press. It can repeat the notification as often as it likes, stack several at once, or swap the name on an existing one.
The notification wears the trusted app’s own name and icon. Nothing about it points to the app or attacker that actually sent it.
The reminder isn’t the only door. In the last build we analyzed, the same manifest has several other receivers open to any app, with no permission on them:
Why it matters
Callback scams work because the victim places the call. Nobody screens an outgoing call.
“HR called you an hour ago” is hard to ignore. So is “IT Service Desk” or a manager’s name. The callback lands on someone who is ready to walk the employee through an MFA reset, read back a code that just arrived by text, confirm a payroll change or approve a payment. Help desk impersonation has been the opening move in some of the largest intrusions of the last few years, and this hands the attacker a credible first touch from inside a trusted app.
The same trick works off the clock. Virtual kidnapping scams already spoof a family member’s caller ID to make the ransom call feel real, and a “missed call from Mom” an hour earlier does that groundwork for them.
The attacking app doesn’t need to look dangerous, either. It asks for no permissions, so it can pass as a flashlight or a game, and an MTD that scores apps on what they request has nothing to flag. It’s not a system controlled permission or something to be caught by a signature check. It’s an undisclosed vulnerability, the type of thing attackers actually use and win.
How we found it
Q-scout analyzes the compiled app the way it ships, with no source code needed. It reads the manifest for exported components and what protects them, then traces what each one does in code. Here it flagged the open reminder component and the other unprotected receivers. We then reproduced the behavior on a physical device to confirm the user-facing result, which is what you see in the screenshots.
What to do
If you manage devices, start checking for what apps are actually doing, not only their published use. Vet communication apps for zero day issues, not just for system permissions and destinations. They have access to critical data and trusted communications paths, they need real analysis not only signature checks and “Top 10 Mappings”.
If you’re a user, a missed-call notification is not proof anyone called. Call people back from your contacts or the company directory, not from a button in a notification.
Quokka can tell you which app this is and whether it’s on your fleet.
Contact Quokka to learn which app this is and to see what the apps on your devices expose to each other.
