Really who created these safety filters

Google AI Studio
, ,
26

Now i tested it with some shiva mantra it created that filter specifically blocks shakti mantras really are you guys serious

27

Still blocks something about shakti well i can bare not talking about 2T physics maybe but this is redline for me censorship on shakti is untolerable i know flow things are not about ai studio but there is nowhere except here that i can give real feedbacks about it anyways

28

I tried same mantra in gemini app it blocks too

And on ai studio lyria 3 blocks too so it is completely relevant to ai studio too

29

I analyzed which flag got triggered and it was copyright flag so it means as i understood the name of the goddess can be owned under copyright it is really interesting

30

Ever since the days of Bard, the early GPTs and Claude, when models would block my content because I asked how wide my stance should be for a squat at my height to avoid injury, I’ve had a deep aversion to any kind of content censorship by AI.

Whoever set up those filters in AI Studio did it deliberately and out of spite. Because the exact same content goes through without any issues on https://gemini.google.com/. So someone intentionally wanted to block the majority of ordinary tasks in AI Studio.

31

Guys, let’s have faith that Gemini 3.5 Pro will be the ‘2.5 Pro’ out of respect for the generational leap, and that it will be just like it—more free, without the bias that comes from trying to apply filters, etc. I know the Google AI team actually reads these comments (even if they don’t want to reply). I don’t know what the barrier is to making the model more ‘accessible’ when:

  • Since you’re on a paid plan, you’re an adult; they should leave the models more ‘free’ if you’re on a paid subscription, as this confirms you are an adult and it falls under your legal responsibility.

  • Man, if they don’t want the Gemini app to have those models, then leave them in AI Studio (an average user doesn’t know about it by a long shot).

  • They should add a filter toggle option to turn them off (only strengthening the guardrails to prevent the AI from helping with hacking, overly illegal stuff, etc.).

  • Leave the filtered model for regular users who don’t have a paid Gemini plan (that way they save themselves the headache of figuring out who is an adult and who isn’t).

  • I know that the higher-ups obviously have to make the decisions, but they should still think about the customer’s needs. We aren’t asking for the wonder of the world; we are asking for a beneficial adjustment for both sides. You win and we win—it’s a win-win.

32

At the very least, the Gemini team has no intention of changing their approach; their content filters have become even stricter today.

Previously, I was flagged by the content moderation system for discussing a plot involving Cthulhu elements, and it took three days for my account to be restored. Today, I wasn’t even discussing any real-world plotlines—I simply posted a character design sketch (with absolutely no adult content)—and it was still blocked.

33

But this is not a model filter, it’s an AI Studio filter. Because the same text goes through on the app without any issues. So it was intentionally introduced in AI Studio with some purpose.