Please check this other discussion on the same subject, and especially my post : What about an AI badge in Storeman - #11 by Seo
How you place your trust is your own choice, but other people have reasons to trust differently. Hence the need to be transparent about the kind of help provided by the LLM tools.
You didn’t address the other requirements (formal proof of people writing software being actually competent).
Why would I ? I don’t think you can prove that part at all, paying good money, or trusting certifications does not prove competence.
For that part my offer was a quality control process and sand boxing.
Because according to you the “house analogy” is correct. Well, it is not: there is key difference between physical objects and software, especially small apps running on user’s phone; apps that rarely carry any option of harming users (unless users are brain-defficient enough to use software of unknown origin to do stuff that might harm them).
But you at least can provide a list of formal requirements and related things, right? Otherwise the app itself is simply “vibed” (you add features as you go with no proper architectural considerations, consequence analysis, etc.).
Yeah I didn’t expand on this in my post, but I agree that there should be some granularity to the disclosure. It’s not black and white.
The whole point of trust is to not need to disclose/prove absolutely everything. Building trust with people online is much more difficult than in person, but it’s possible. I trust many devs in the community. But I don’t blindly trust LLMs.
Yeah this is another point with nuance. Like there are different kinds of buildings, there are different kinds of apps. A sailjailed app with no permissions should be quite low-risk, while a browser is a whole different story.
Thank you for your answer. You know for yourself that the workflow/requirements you mention are way above what is being demanded here - you might need your full list to avoid being sued into oblivion for selling a product that large firms or governments are about to use and that allegedly doesn’t meet the standards that were agreed on in hundreds of jour-fixes, meetings and kick-offs. Also, this list reflects very much how large funds are needed to be able to produce such kind of secure software, and the revenue that can be expected should the buyer agree to purchase this product.
A SFOS app is neither worth billions of money, nor does life itself depend on it, nor is it used by more than 200.000 users. It doesn’t come with meetings, cost allocation supervision, daily scrum standups, it isn’t ordered and paid, isn’t tested externally and automatedly, and in the end noone is being sued or rewarded for bad or good work.
All that is being asked for is how it has been made/if there are tools involved that create additional risks, and AI cannot be held liable.
Cheers
I refer to my previous reply:
I don’t really care.
Is the app useful/dangerous to [me]?
And what is being demanded here? There are requirements for Jolla Store apps and for Chum apps, but for OpenRepos? Nope.
What I understood is “you should do XYZ if you want me to use your app!” Great, but I do not care if you use it or not.
You yourself noticed it:
Therefore, why would anybody care enough to bash and brigade people who vibecode stuff and allow others to use it? To me, it looks like a peculiar kind of self-appointed elitism: treating a tiny hobbyist ecosystem like an exclusive club, then punching down at newcomers for not meeting standards that nobody actually imposed or agreed on.
192 people responded, which is 0,096% of alleged user base 
It is not about bullying newcomers, it is about a minimum ruleset. The discussion is moot, as sooner or later the EU AI Act will require every product to have the “made with/by AI” label that has been created with AI means anyway.
It is not “I will only use your app if you did xyz”. It is born out of the “why does this particular flashlight app cause so much network traffic” we had some years ago.
On the other hand, in a not so distant future, most of the useless social media etc. apps will use AI in a way or another. It could even be that a Software Developer who does not know how to use AI to speed up his work, may have difficulties to find or keep his job. I dont mean full vibecoding, but being more selective, i.e. concentrate on the fun creative work, not rewriting the boring stuff that has been written million times already.
While this might have a lot of truth in it, your post makes the distinction between “AI as a controlled tool” and “use AI because I don’t know how to do stuff”. I use algorithms to avoid doing the same stuff all over again day by day, but I will not let that take over the important aspects of my job. AI != person, and only a person can be held liable. People produce outcomes they need to be able to explain to their customers/higher ups, and the insides of AI blackboxes are rather unknown. Tokens get more and more expensive, and before I have to prompt and prompt again and again only to make AI understand what I want or need I just code it myself.
I know AI lover CEO talk (one of mine is exactly like that), and I know that not everyone is convinced enough to follow that, me included.
Cheers
I agree of course, many times it is more effecient to do it yourself. Also here we are mostly talking about hobby projects and time that is “free” anyways. Will be interesting to see how those AI pushing CEOs will react if their software would not get the official “Made with AI” badge 
As addition: Most Building Codes are written in blood.
Of course not, as long as those “newcomers” agree to being bullied and openly insulted by those being here before.
Is there “you must use my app” thingy anywhere? Why would you even use an app that has or does not have “made with LLM™” label? Non-LLM powered apps can also be malicious, as per your own example. For me “vibecoded app” = an interesting experiment solving a particular problem, maybe worth looking into whether time investment might be beneficial”. Heck, even the author of “vibecoding” term used it like this.
Of course it does, because those are two separate things. Yet people here seem to go black/white only.
Simply put: it is up to the end user if he/she/it wants to use given piece of code.
I’ll try answer about the quote: ‘‘they are happy with the current state and would want things to continue this way for another 20 years.’’
The thing is many old users, me included, are still absolutely enchanted by the idea that there actually ‘is’ now a mostly-woking alternative to the 2 Giant players in phone software. Currently exept SFOS, there is no real daily-drivable viable alternative if you do not want iOS or Android (including Android Forks).
I am still absolutely amazed by the fact it exists and didnt got thrown under the bus (RiP BB10).
Discovering SFOS feels the same as when someone discovers that linux on a laptop ‘actually works, and even superbly works’ and that Windows and MacOS isnt the ‘only choice’. Its a ‘‘Waoooo’’ feeling.
So some of us are under the spell of this ‘‘Waoo feeling’’ both with linux on laptop and SFOS on mobile. So we are okay with some frictions and shortcoming (tbh there is less and less each day).
Back to topic: So old users like me can be therefore little (lot?) worried by AI since sometimes it feels it could bring more issues than solutions. Like vibe coded apps ruining battery life even after closure recently spotted in OpenRepos. We have been mostly happy with human written code for years, so we are worried that issues could be bigger to swallow than the ‘‘developpement-speed kick’’ it can offer.
To conclude on that: This is the whole debate about the global use of AI overall isn’t it? Like is it a helping hand, or something unsafe that is getting soon out of human control?
I’ve made contributions (human) to several LLM coded (claude) projects. I thought they were fun and the code rather good. Zero responses from the developers (over a period of some months. They pushed to open repos and disappeared. I will not contribute to one of those projects again. I did have a great experience with the developer of FinTube, @masen who is not really ‘vibe-coding’ but is not only responsive to feedback, but often faster than the PR machine (me) trying to contribute. So, it’s not black and white. Sorting it out is a bit messy.
I have been working on something for some time now. A camera app that uses a Camera2 api bridge independent of droidmedia(only device initialization/de is called). I used LLM for the breakthrough in the pipeline and interpolation algorithms. I have been reluctant to share anything publicly about it at this stage to avoid controversy and debate over the use of LLM rather than on the actual project; even though I am using LLMs as a research and assisting tool rather than a full blown(do what I say model). The way I think of LLM is akin to hammer; a tool no more no less. You either build or you demolish with it. The human using it dictates the direction and choice. I would perhaps assume that people are leaving projects due to LLM fatigue or some sort of instant gratification deficit that accompanies the rapid nature of LLM use. Reading an obscure documentation and building something with it grants a sense of accomplishment that LLM use will never be able to replicate.