Google News Scraper — LLM-Ready Articles by Keyword & Topic

Scrape Google News for any keyword, topic section, or the day's top headlines — in any country and language — and get clean, structured articles: title, source, publish date, snippet, and the real publisher URL. Optionally pull the full article text, ready to drop into ChatGPT, Claude, or a RAG pipeline.

Everything comes from Google News's official RSS feeds — the same feeds Google publishes for anyone to consume. No API key, no headless browser, no proxies. That keeps runs fast, cheap, and reliable.

Why this one

Most Google News scrapers hand you the raw feed and stop. Two things here don't:

• Real publisher URLs. Google News links are opaque redirects like news.google.com/rss/articles/CBMi…. This Actor resolves each one to the actual article URL on the publisher's site, so the data is usable in a database or a citation — not a dead-end redirect.
• LLM-ready full text. Turn on Extract full article text and each article is followed and cleaned into plain body text for summaries, sentiment analysis, and retrieval pipelines.

What it does

• Search Google News with multiple queries at once, including Google operators (site:reuters.com, "exact phrase", intitle:, when:7d).
• Pull topic sections: World, Nation, Business, Technology, Entertainment, Sports, Science, Health.
• Fetch the country's Top headlines feed.
• Any country + language edition (US, GB, AE, IN, ES, FR, DE, and more).
• Recency filter — past hour / day / week / month / year.
• Deduplicate stories that surface across several feeds.
• Export to JSON, CSV, or Excel, or pull via the Apify API.

Input

Field	Description
queries	Search queries. Supports Google News operators.
topics	Topic sections to include (Business, Technology, …).
topHeadlines	Also fetch the main "Top stories" feed.
country / language	Two-letter codes for the edition, e.g. US / en, AE / ar.
dateFilter	Restrict search results to a recent window.
maxArticlesPerQuery	Cap per feed (Google News RSS returns up to ~100).
decodeUrls	Resolve real publisher URLs (default on).
extractFullText	Follow each article and extract clean body text.
fullTextMaxChars	Truncate extracted text.
deduplicate	Drop repeats (same title + source).

Example

{
  "queries": ["artificial intelligence regulation", "\"interest rates\" site:reuters.com"],
  "topics": ["BUSINESS", "TECHNOLOGY"],
  "country": "US",
  "language": "en",
  "dateFilter": "7d",
  "maxArticlesPerQuery": 50,
  "decodeUrls": true,
  "extractFullText": true
}

Output

Each article is one dataset record:

{
  "title": "Designing physics experiments with artificial intelligence",
  "source": "Nature",
  "publishedAt": "2026-07-12T06:35:18+00:00",
  "url": "https://www.nature.com/articles/...",
  "googleNewsUrl": "https://news.google.com/rss/articles/CBMiX0FV...",
  "snippet": "Researchers used machine learning to...",
  "query": "artificial intelligence regulation",
  "topic": null,
  "feedType": "search",
  "language": "en-US",
  "country": "US",
  "fullText": "Full cleaned article body text..."
}

Common uses

• Brand & competitor monitoring — schedule a query per brand and get a daily feed.
• Market & sentiment research — full-text articles into an LLM for tone and theme extraction.
• RAG / knowledge bases — keep an index fresh with clean, deduplicated news text.
• Newsletters & dashboards — topic and headline feeds, normalized and export-ready.

Notes

• Full-text extraction depends on each publisher's page and works best on standard article pages; paywalled or heavily scripted sites may return little or no body text (the article is still included with its metadata).
• URL decoding is best-effort — if Google changes its redirect scheme for a given item, the record keeps the original googleNewsUrl so you never lose the article.
• This Actor reads only public RSS feeds and public article pages.