I'm firmly on the side of "AI" skepticism, but even I have to admit that this is a very good use of the tech. LLMs generally do a great job at summarizing text, which is essentially what this is. The sources could be statically defined in advance, given that they know where they pull the information from, so I don't think the LLM generates that content.
So if this automates the process of fetching the top news from a static list of news sites and summarizing the content in a specific structure, there's not much that can go wrong there. There's a very small chance that the LLM would hallucinate when asked to summarize a relatively short amount of text.