Internet Archive
The Internet Archive is a nonprofit digital library that has been preserving digital and digitised culture since 1996. Its collections run to millions of items: scanned books and periodicals, audio and music, film and video, television news, vintage software, and the Wayback Machine, which has saved more than 800 billion web pages and lets you view a site as it looked at a chosen moment in the past.
Most items can be browsed and downloaded straight from the web interface, and there is a rich set of programmatic options behind it. The Archive offers metadata and search APIs, per-item download endpoints, and the community "internetarchive" command-line tool and Python library for fetching items and their metadata in bulk. The Wayback Machine has its own APIs, including the CDX index, for finding and retrieving historical page captures by URL and date.
For RAG, the Archive is most valuable when you need text that is not easily found live: out-of-print and public domain books, historical periodicals, archived versions of web pages that have since changed or vanished, and transcripts from its television news collection. The Wayback Machine in particular is the go-to source when a question depends on how a page read on a specific date, which matters for anything time-sensitive or evidential.
The big caveat is licensing. The Archive is a library, not a single-licence dataset: material ranges from clearly public domain works to in-copyright items made available under controlled digital lending, and its overall reuse terms are best treated as unclear. Always check the rights statement on each collection or item before you reuse the content commercially, rather than assuming everything is free to build on. OCR quality on scanned books also varies, so expect some noise in older texts, and be considerate with bulk requests, since it is a nonprofit running on donations. It complements Common Crawl (live web at scale) and Project Gutenberg (clean public domain e-books) by covering the historical and out-of-print material those miss.
Related sources
CC-News
A subset of Common Crawl focused on news, containing millions of articles pulled from news websites around the world. It has fed several large language model training pipelines and gives you a ready news corpus without crawling sites yourself.
GDELT
GDELT, the Global Database of Events, Language, and Tone, monitors print, broadcast, and web news in more than 100 languages from every country. It records events back to 1979, builds a Global Knowledge Graph, and adds sentiment and emotion analysis, all queryable on Google BigQuery.