The underlying NotebookLM is doing better at this - each claim in the note cites a block of text in the source. So it’s engineered to be more factually grounded

The underlying NotebookLM is doing better at this - each claim in the note cites a block of text in the source. So it’s engineered to be more factually grounded. I would not be surprised if the second pass to generate the podcast style loses some of this fidelity.