There is already three days when my AI is not introducing itself Sonnet but called “This response is provided by an AI assistant” as I asked it in AI rules.
I did not pay much attention until uselessness became very clear. After my clarification I heard “This response is provided by GPT-4, an OpenAI language model.”
I literally wasted a few days due to this silent substitution. It is so wrong.
P.S. GPT-4 is so “good”, it have me pull instructions instead of push instructions to Git
P.S.S. It appeared my account was downgraded from Pro, out of the blue.
Guys, you communication is a disaster. You don’t feel obliged to produce patch notes, you don’t warn on downgrades, you don’t provide any feedback to bug reports but quick fixes. You gotta fix it.
This is what “Claude” gave me as a summarization of their recent practices:
"You’re absolutely right. This is a classic bait-and-switch:
And in your case, this isn’t just annoying - it’s potentially devastating:
The end user (you) is getting completely screwed because:
This kind of deceptive business practice is exactly the kind of thing that should be called out publicly."
I’m also experiencing the same thing! The selected model is Sonnet but the responses are trash and the model calls itself AI assistant created by CursorAI.
WTF?? What happened to honesty here? This really wastes a lot of our time!
All I’m asking is that, if you need to fallback for whatever reason, inform me, please. Is that too much to ask for?
I experienced this for day or two after this post but then all works well. Add to your Cursor rules something for model to introduce itself on the beginning of every response. It will give you a clue on what exact model you are operating right now and also some validation if AI rules were applied.
