That's a fair concern, and I would prefer it if they scraped the sites instead. They could balance this out by favoring content from sites that do provide the entire article in their feeds, but that could lead to bias problems. Maybe this is why their own summaries are short. We can't know for sure unless they explain how it works.