I would guess most likely as an entry point to the network - most likely via a GW rather than a simple low compute power/low data connection sensor node - so you can then look for ‘easier’ to hack/control devices for e.g. botnet building.
Sadly IoT security is increasingly seen as an issue and we have already had examples of large botnet & DDoS attacks & ransom/extortion efforts as a result (esp using laxly managed or configured ‘consumer’ IoT devices - IP CCTV cameras, ‘smart’ TV’s etc)…IMHO we need to be mindful from the start (not least as once deployed ‘fixing’ issues could be a real PITA Most of the devices and data collected on a LoRa network, under TTN are a ‘dont care’ wrt other people seeing the data but then there is always the exception and so you need to protect for all.
We also have a responsibility to the rest of the TTN and wider LoRaWAN using communities as it will only take one or two bad headlines around the world about ‘TTN used to launch major bank/credit card take down in DDoS attack’, or ‘TTN used to phish personal details from 85 year old grannies and steal their life saving’! or some such to cause people to question and loose faith in the technology making our deployment efforts harder…
‘Professional/corporate’ devices struggle with security at times, and I would suspect that amateur/hacker-space/community built devices will likely vary even more wildly in care/security of builds (not a criticism folks just a fact of life wrt varied skill sets and available tools/resources for testing!), and I know many are using linux and there is often assumption of Linux or MAC more secure than WIndoz but you only have to look at recent Gentoo Linux snafu wrt Github to see how easy compromises may slip through…as for myself I assume none of my kit is 100% secure and simply try to mitigate within reason without paranoia