I am wondering how SPV handles the following scenario:

1. A transaction has been confirmed by some block X, and funds address y.
2. At some later block, the transaction's output are spent again
3. SPV node queries the balances of y, the full node peer returns the answer along with the merkle proof for block X.
4. But.. the answer doesn't give you the most up to date response, because y funds were depleted in a more recent block...

I know some ideas like UTXO commitments were around but I am not sure that they were ever merged into bitcoin.

So, was this problem ever solved? Or maybe no one tended to it because it was concluded to not be so bad?