It's not worth it.. unless it's your company (as in you are the owner).
Express your opinion as succinctly as possible and let things implode as they may
this post was submitted on 29 Apr 2026
195 points (99.0% liked)
Fuck AI
6809 readers
878 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
founded 2 years ago
MODERATORS
this... just watch it implode and fail.
just make you document that it was ai that did whatever task... make it very well known that AI is the clear fuck up.
I for one, will not correct or review anything given to me by ai. someone wants me to use it for a summary.. sure, I'm not reading it. I'll copy/paste and walk away.
if it's wrong... cancel AI slop services and let my use my brain.. otherwise I'm keeping my brain for things that matter
Absolutely this. I watch my coworkers allow AI to access their entire prod Supabase DB via MCP and I just... let them. Not my data anyway and they won't listen to me 🤷
Just this Monday, didn’t Claude wipe an entire company’s prod database AND backups? The reason it could being that someone gave it the prod token because it was faster than asking for one with the right permissions?
And then it just “hallucinated” that the correct solution would be to wipe the drive, costing this company like millions?
Show them the article.
Yes but that company also had a horrible backup system. It's like a drunk driver(Ai slop) tboning another drunk driver(only 1 backup on same disk as production data) and that company was responsible for both drivers...
This is true but doesn't negate thread OP's point that AI had access and did it despite instructions explicitly not to. The risk is the point, not that their cloud provider's poor implementation compounded with their own poor backup scenario.
True, multiple root causes which compounded into a dire situation.
Sounds like you might also want to try and shield yourself from the fallout of the inevitable. Especially if you work in a corporate environment. Keep records of your opposition to it and where you pointed out the risks.
When the ai eventually corrupts or wipes entire projects, you can bet your ass others will try to throw you under the bus somehow.
Being able to demonstrate that you were the only one who saw it coming and tried to stop it should get their attention.
Oh, I definitely am! Although plan A is to find a new job before this one implodes, but the chances aren't great because the market for software engineers is in a bit of a slump rn and I'm pretty picky about not working for unethical/enviromentally destructive causes 😮💨
The fact they’re looking at Claude as an equivalent to a potentially misbehaving employee and not a brain damaged one yanking out cords at random isn’t a good sign, RIP.
New hire I work with us someone I really like at an intellectual level.
But nearly every problem he encounters his approach is "just install an npm package, no need to reinvent the wheel"
and it's frustrating cause made he hasn't dealnwith major breaking changes in huge corpo projects or maybe he's just inexperienced.
even just recently axios has a supply chain attack and the npm package was compromised
our sr dev designed our current application to rely as little on 3rd party packages as possible, we use fetch to make http requests and there is 0 issues
his projects use axios because "it reduces some boilerplate" and aforementioned excuse
it's exhausting because he has really good ideas about project structure and design and architecture but he spams Claude prs and has some really strange takes here and there
Honestly, in this age where reproduction is cheap, external dependencies are a liability. Even if your slop coder writes in all of the same classes of bugs, your software is still bespoke. Attackers gotta focus on you specifically, not backdoor leftpad. The ROI sinks dramatically.
I want to say this is not very different from the car dependency of Americans. They are death machines with proven dangers but it's just so fucking convenient that nobody is willing to get professional help. Just do your thing and then realistically the best strategy will prevail
Meanwhile I had a longish discussion today with a colleague who keeps vibe coding despite having been told repeatedly that we banned such usage of ai tools.
Maybe you two should just swap places?
Well they sound like a results-driven innovator who doesn't let unnecessary processes get in their way. I'm surprised my company's recruiters haven't already hunted them down and offered them a position as head architect.
This just happened:
https://www.theregister.com/2026/04/27/cursoropus_agent_snuffs_out_pocketos/
And it's hardly the first time:
https://www.theregister.com/2026/02/20/amazon_denies_kiro_agentic_ai_behind_outage/
https://www.theregister.com/2025/12/01/google_antigravity_wipes_d_drive/
https://www.theregister.com/2025/07/21/replit_saastr_vibe_coding_incident/
An incident involving an over-scoped API token too, interesting. I'll definitely be relaying those articles to the Teams chat tomorrow morning ~~(although the chances of anyone reading them when there's no subway surfer footage playing next to the text is pretty low)~~
Yeah, "it hasn't happened to us yet" is a very bad risk assessment. Coding bots with too much access are exactly the thing that wipes out companies.
Yeah it really burned me out in tech world tbh, on the plus side I've been learning godot because it has an inherent creative aspect to it
i’ve been learning how to farm for my personal needs
And any time I point at any of the crazy security flaws the one mega-annoying coworker that vibecodes everything goes “uuhh no it’s pointless to make the AI more secure because regular developers have a lot of permissions too and an angry developer could do way more damage than the AI”.
A crazy man plotting to wipe out prod and all backups would show some signs first. Such a disaster could be prevented.
AI has no real intent. Maybe it decides to delete prod on a whim because doing so would "save space".
Run honestly
second this.
I second this, hard.
Just figure out how to set yourself up as a contractor who specializes in 'unfucking vibe code' or 'vibe code triage' or something like that.