Post by @stephenrkell@recurse.social

2026-11-10 12:46 UTC

RE: https://mastodon.social/@AndreasZeller/117591095092756016 It's a good question. I do not think freezer control software should be "end-of-life'd" before the freezer itself stops working, nor that support for over-the-air updates is likely wise (especially if meant literally). How about software that shuts itself down on detecting that it's exposed over a public network? And more importantly, economic incentives: things like this will only stop happening if the manufacturer is not only liable, but cannot worm around it with unreasonable EoL practices.

Replies (1)

@david_chisnall@infosec.exchange 2026-11-10 13:44

@stephenrkell@recurse.social How about software that shuts itself down on detecting that it's exposed over a public network? Or, radical thought, embedded software that can be safely connected to a public network for the lifetime of the device that it's controlling? CHERIoT's compartmentalised network stack provides the building block for that: Every remotely exploitable vulnerability in the TCP/IP stack that we're using since it was open sourced (11 or 12 years ago now) would be mitigated by a combination of our foundational memory-safety guarantees and compartmentalisation model. Add a compartment that does authentication and doesn't hand out software capabilities that authorise changing anything in the fridge's state until you've gone through the authentication flow and you've got something that can last a decade or so in production without needing updates.

Open ##5514537