# Should I just stop worrying and trust my AI? — retained record

> Retained reddit record from reddit:selfhosted.

Canonical: https://highsignal.app/data/records/da3ea74c0e0bf44f

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:selfhosted)  published 2026-10-23 01:59 ingested 2026-10-24 02:38

[original source ↗](https://www.reddit.com/r/selfhosted/comments/1xd1dft/should_i_just_stop_worrying_and_trust_my_ai/)

## retained text



```
Everyone making such a big deal about security of AI agents, but does it really apply to me?

I have a homelab guy with a VPS and two small servers. There's nothing of value saved anywhere, everything is backed up to git or my NAS and can be recovered. I just use it for personal projects.

I'm running an AI assistant in Docker, it's a minimal alternative to OpenClaw called Nanobot. I use this assistant for various IT tasks. I have it customize my docker compose files, vibecode apps, port x64 apps to my ARM server, troubleshoot general Linux issues (for example earlier today xrdp wasn't working).

So far I've been treating the AI as untrusted. I review the code/scripts it writes. I don't give it SSH access to my server. I have it spoonfeed me troubleshooting commands and I waste my time reviewing and copy-pasting them.

In the 2 years of using AI I've never had one tell me to do anything malicious. My assistant isn't visiting random websites likely to hijack it, assuming such a thing is even likely to happen. When it does a web search, it's usually looking up specific tech doc, for example how to use some Go library on Github.

I feel like if I just say "fuck it" and let go of the paranoia, my homelab life would become 10x simpler and I'd win back a lot of free time. And if my server was somehow hacked (which I don't think it will), well whatever, I'll format, reinstall, recover from backup, and lesson learned. I bet that would take less time than what I've been doing so far.

Should I just stop caring and free up my time? Have you given your AI assistant SSH access to your servers?

I know this sub will bury anything AI related and I'm gonna stay at 0 downvotes within a minute of clicking Submit, but I'm hoping a few people who have done the same will see this thread and can tell me how that went for them.
```



document fetched 2026-10-24 00:30 · canonical https://reddit.com/r/selfhosted/comments/1xd1dft/should_i_just_stop_worrying_and_trust_my_ai
