It said she didn't have a cloud subscription, but that there are data pipelines that make these sort of devices work. (Imagine there's a thumbnail of the video in the product somewhere, so there's a pipeline that takes a video stream and generates thumbnails.)
According to the article, it was a matter of having someone figure out which pipelines her videos might have touched, and then go looking to see if there were any ephemeral artifacts that hadn't been lost yet.
https://www.cnn.com/2026/02/10/tech/google-video-nancy-guthr... and archived at https://archive.ph/oZyRM
No consumer product should have users do port-forwarding or punch holes in the firewall. You don't want an IoT device on your network accepting packets from the internet.
The proper way to do this is with a cloud server arbitrating connections, which is what a lot of products do.
The reason most consumers want cloud storage isn't for ease of access, though. It's because they want the footage stored securely somewhere. If the thief can just pick up your camera and walk away with the evidence, it's not very useful to you.
They're not architecturally delivering the video a different way if you pay than if you don't. They're just changing the retention period.
This video was probably recovered from cache somewhere.