Which is different from believing the information on an ad and dark pattern infected website how?

Since the AI is trained on the data on the websites below the AI summary, the summary quality is basically lock step with the quality of the websites.

It is different because the AI is 'Google' branded and not 'ad and dark pattern infected website' branded. That proximity to Google's branding conveys trust, but it just whitewashes the content from that ad and dark pattern infected website. Only now you don't know about it.