False Positive Trigger / Context-Blindness in Safety Layer (Ref: Gemini 3 Flash)

[Truans]

Hello! False filter triggers are a very common problem right now. If you have screenshots with examples, I recommend sending them in response to Logan’s comment:

Did Google AI Studio silently change safety filtering today? Full responses now get erased instead of stopping generation – this breaks creative writing and RP

Hey folks, I have not seen this change, does anyone have a screen recording of the behavior? Generally, AI Studio is not a creative writing platform. The playground is meant as a place to quickly test Google’s models as a developer, then get an API key and go build something with it. I think for developer use cases, you can use the API how you were initially describing, but not sure about this safety change, in general it seems to be working as intended but happy to see the screenshot / recordin…