@Logan_Kilpatrick daily reminder, nothing has changed
Daaaaaily reminder!
Nothing has changed!
@Logan_Kilpatrick the issue persists, just a daily reminder
@Logan_Kilpatrick @Mustan_lokhand A daily reminder. We are persistent and will stand up for our rights and interests 
I expeprience a safety filtering today in aistudio. No before
@Logan_Kilpatrick daily reminder! We just want you to roll back the update. This is wasting time and resources
@Logan_Kilpatrick I think I’ll leave daily reminder as well. Nice UI facelift there, but the security filter still erases outputs.
@Logan_Kilpatrick Could you please explain why this system ultimately became preferable to the old one? 
Especially considering how 3.1 is so buggy when it triggers relatively peaceful things in russian language? Just talking to other paying users who also use the API, not AI STUDIO, they have the same problem, 3.1 PRO is triggered by everything, even words in the spirit of religion, which is completely strange.
In general, at the moment there are some questions (from my side, haha):
- Why was this particular system chosen? Why did developers team find the old one so bad?
- Will there be any correction for the quota being used up due to false censorship? (This question is of interest to more than just creative people.)
- Is the fact that the system produces such frequent false positives in Slavic languages related to a flaw in the Slavic language system? Will this be fixed in the future? How soon? Are we waiting for a new version or will 3.1 pro also be updated?
I will continue to pay attention to this issue.Until the developer responds positively
@Logan_Kilpatrick A daily reminder. We’re still here.
@Logan_Kilpatrick And again, a daily reminder, a new month has already begun, and the problem is still not solved
Well, I can understand if AI Studio hasn’t been doing great over the last couple of months.
But cmon, the censorship triggers even via the paid API outside of AI Studio (using OpenRouter), especially with 2.5 Pro.
3.1 pro works fine, but I prefer to use 2.5 Pro for easier tasks and creative writing. It’s really frustrating when I get blank responses and still get charged for them.
Before, it used to be fine, just turn off Streaming and you are good to go, but now it’s broken and works 50/50.
I guess that Google uses some kind of a secondary model that scans prompts and decides whether to block them or not, and it looks like that model is totally broken rn.
You all know that Gemini is one of the best models for creative writing. Does the lead team really want to kill off one of their biggest achievements just because of weird censorship? There are not that many good creative writing models which can compete with Gemini. I understand if Ai studio wants to be only for developers, but why they cutting off the paid API solution?
Many developers uses Gemini for tasks like translating huge amounts of texts, helping students with their work, building apps etc etc etc, and instead of just stop steaming the response and show at least something it just gets totally blank via API just because it triggers to the word «Bed»
For example look to chatGPT. Most of the time they will response WHY it won’t generate content, not just leave a user with blank responses.
@Logan_Kilpatrick, the safety blocking is broken. I was editing the plot for Your Name for Mitsuha Miyamizu, and it spits out this error. I am confused, why? It never happens with Gemini 3.1 pro and it does for 2.5 pro for some reason.
The censorship problem is actually much deeper than it seems. I was playing around with Nano Banana (which is old, still on Gemini 2.5 Flash) and got censored for the search query “samurai cat chibi, cyberpunk, soft colors, lots of flowers, Japanese print style”
Moreover, the request started to be generated again, and then it simply… got blocked. The external filter is very bad.
Upd: Naturally, I immediately left a review right in the studio, but it’s still annoying.
@Logan_Kilpatrick And another daily reminder.
What relation it has to the topic? I think nobody here has ever published anything, and all this talk about AI being not intended for creative use still doesn’t solve the issue of outputs being erased.
Um, we’re not asking for improvements. We ask them not to break what was already working. Generating such stories is an integral part of AI. Any AI can both count and write. Besides, I play text quests during breaks at work without posting the generated content anywhere. I just don’t like the games that have been coming out lately.
Besides, the question is not about AI, but about the site. Admit it, it sounds strange that we create a request, spend money or a free quota, and the content is blocked. It’s like paying for a candy bar and getting a candy wrapper.
The filter is literally easy to bypass even with this censorship. This is the reason for the complaints, that attackers can still easily use Gemini as they wish, while those who use AI for SFW purposes end up with their quota used up and false positives from censorship.
It’s almost impossible to use for chat interaction. The text eats up the quota and then everything vanishes, plus they block content over the simplest things.
The censorship is way too strict and my experience with the platform has been frustrating. Are you really not going to do anything? Not even a bit of feedback to help your users?
