I agree. I tried reading some of the documents and they're full of this:

> LIMITATION: Direct PDF downloads returned 403 errors. ProPublica Schedule I viewer loads data dynamically (JavaScript), preventing extraction via WebFetch. The 2024 public disclosure copy on sixteenthirtyfund.org was also blocked.

> Tech Transparency Project report: The article "Inside Meta's Spin Machine on Kids and Social Media" at techtransparencyproject.org likely contains detailed ConnectSafely/Meta funding analysis but was blocked (403)

The least they could have done is read their own reports and then provided the documents to the LLM. Instead they just let it run and propose connections, asked it to generate some graphs, and then hit publish.