Yeah, its not unreasonable that you’d have a remote way to access the device to gather debug data with the customers consent. An SSH key in the firmware is a flexible way to do that, so long as there are good controls in place to ensure that it isn’t misused.

I think multiple people already have access to the databases that store the data the device sends. I don’t really care whether they get the data from the device itself or from the database.

Similarly, I think multiple people have the ability to make changes to the firmware build and the systems that distribute it. So those people already have the potential ability to gain access to the device.

One person or multiple people having unauthorised access are both unacceptable. I’m saying that the users have to trust the companies ability to prevent that occurring, and that therefore this particular technical detail is mostly irrelevant

I’m 90% sure it is not a single user. I just don’t see how that really affects the security of the product, given that the company that sells it can already do the things the author is saying can be done if you have this key.

To be clear, I wouldn’t buy this. I just don’t think the SSH key makes it any worse than it already was

A shared account doesn’t mean everyone who works there has access to it, or that those who do have access aren’t subject to some type of access control.

The article basically goes on to say that the existence of this key makes a huge difference to the security/privacy of the product. It argues that using it, someone could access data from the device, or use it to upload arbitrary code to the device for it to run. However, those are both things the user is already trusting the company with. They have to trust that the company has access controls/policies to prevent individual rogue employees doing the things described. It seems unreasonable to say that an SSH key being on the device demonstrates that those controls aren’t in place.

The email address attached to the public key, eng@eightsleep.com, to me suggests the private key is likely accessible to the entire engineering team.

This assumption is doing a lot of heavy lifting in the authors argument that this is a big deal.