Six months of coordinated disclosure on a TP-Link Kasa camera resulted in two CVEs, a triage failure where the vendor described a vulnerability that doesn't exist in the reported payload, a beta patch that permanently bricked my test device, and a factory reset that doesn't clear previous owner data.
The GPS finding (CVE-2026-13230) has been publicly documented on this device class since 2020. A single UDP packet returns sub-meter home coordinates with no authentication required. TP-Link scored it 5.3 medium. My independent assessment is 7.1 high. Precise home coordinates aren't low confidentiality impact.
The credential finding (CVE-2026-9770) covers a fleet wide RSA key and unsalted MD5 TP-Link ID credentials. Same credentials provide global authentication across the TP-Link ecosystem.
Factory reset on a secondhand device doesn't clear the data. Connecting to the device's soft AP during setup and sending a single UDP packet returns the previous owner's GPS coordinates.
The report seems obviously AI generated, so I can't be bothered to read in its entirety, but based on my quick skim, "leaked home GPS" makes it sound worse than it is. Unless you're dumb enough to set DMZ on this device, this won't be exposed to the internet, and if it's LAN only, don't you already know the location? Even for a remote attacker who somehow got LAN access remotely, they can probably deduce the location through other means (eg. using crowdsourced wifi databases).
This underscores the principle that IoT devices should not be allowed to communicate over the public Internet. Pretty much all cheap, Chinese-made hardware of this kind has intentional or unintentional security holes waiting to be exploited.
I think no country can be trusted and therefore all IoT device communication (telemetry/updates) should go through government-controlled local servers in an unecrypted form. This way local government can enforce its policy over foreign manufacturers and has a "kill switch" if they misbehave.
Also, all communication and telemetry should be opt-in and turned off by default.
Government-controlled server would prevent foreign countries from collecting intelligence and pushing malicious updates.
A shocking number of devices are continuously reporting location data over random unencrypted protocols. What’s worse, they’re often sending the data to cloud IPs that aren’t even controlled by the company, so some random person is getting your real-time location.
I've used Kasa plugs for a long while and was not surprised that their API allowed relay control and basic info of them as long as you manage to get in the same internal network. It's local, so IMHO that is not just reasonable, it's desirable. I don't need to give my friends permission to toggle the lights manually either.
Routers having abnormal amount of zerodays, and not being fixed on the other hand is actually serious, unlike this.
Just a week ago I actually set up one of TP-link's new line of smartplugs (Tapo instead of the old Kasa), and for that I had to make an account. For actual security, I'd rather have an option to control them locally with zero additional authentication when you're already inside the network, instead of the cloud stuff. But I HAD to make an account even though the custom code I control said plug with only accesses the plug locally.
Good, they used UDP for a one off request /response (maybe) exchange. Bad, they included precise location information.
I can see how rough location information is helpful for support and business information. Maybe country, maybe even zip code. But precise GPS was a bit overkill. Maybe it was easy, maybe it was nefarious, but not encrypting it over the wow was just plain dumb. I guess there is a razor for that.
How would this exfiltration happen though? Aren’t these cameras going to be behind a firewall? Without a request originating internally no external packet will make it past, right? Does the firmware make the first request? If so, I missed it.
I’m more mystified by the fleet wide certs. Old manufacturing tech that makes per-device firmware difficult, perhaps?
The UDP is through the broken, since 2016, TP link smart home protol. Exfiltration would require a precursor network foothold for a pure network vector, or for local 2nd hand markets the data is returned from the device broadcasted AP which is used for account binding.
15 comments
[ 0.26 ms ] story [ 5.6 ms ] threadThe GPS finding (CVE-2026-13230) has been publicly documented on this device class since 2020. A single UDP packet returns sub-meter home coordinates with no authentication required. TP-Link scored it 5.3 medium. My independent assessment is 7.1 high. Precise home coordinates aren't low confidentiality impact.
The credential finding (CVE-2026-9770) covers a fleet wide RSA key and unsalted MD5 TP-Link ID credentials. Same credentials provide global authentication across the TP-Link ecosystem.
Factory reset on a secondhand device doesn't clear the data. Connecting to the device's soft AP during setup and sending a single UDP packet returns the previous owner's GPS coordinates.
Also, all communication and telemetry should be opt-in and turned off by default.
Government-controlled server would prevent foreign countries from collecting intelligence and pushing malicious updates.
Routers having abnormal amount of zerodays, and not being fixed on the other hand is actually serious, unlike this.
Just a week ago I actually set up one of TP-link's new line of smartplugs (Tapo instead of the old Kasa), and for that I had to make an account. For actual security, I'd rather have an option to control them locally with zero additional authentication when you're already inside the network, instead of the cloud stuff. But I HAD to make an account even though the custom code I control said plug with only accesses the plug locally.
I can see how rough location information is helpful for support and business information. Maybe country, maybe even zip code. But precise GPS was a bit overkill. Maybe it was easy, maybe it was nefarious, but not encrypting it over the wow was just plain dumb. I guess there is a razor for that.
How would this exfiltration happen though? Aren’t these cameras going to be behind a firewall? Without a request originating internally no external packet will make it past, right? Does the firmware make the first request? If so, I missed it.
I’m more mystified by the fleet wide certs. Old manufacturing tech that makes per-device firmware difficult, perhaps?