Google News Scraper & API: keyword search, topics, real article URLs, full text

Get Google News results as clean JSON, CSV or Excel. Search by keyword with every Google News operator, or read topic and local headlines, for any country and language. For each article you get:

• Title, publisher, publish time and the publisher's home page
• The real article URL, decoded from Google's news.google.com redirect link, so you can open, store or crawl the article itself
• Optional page details from the publisher's site: description (snippet), image, author, published time, site name, language and keywords
• Optional full article text with a word count
• Related coverage of the same story, for topic feeds

$2 per 1,000 articles, real URLs included. Page details are $1 per 1,000 extra. There is no start fee and no monthly rental. Articles whose link can't be decoded, and publisher pages that are blocked or paywalled, are free.

Why use this Google News API

• Built on Google's official RSS feeds. Search, topic and location feeds are the most stable way to read Google News: no headless browser, no HTML parsing of the Google News app. A 50-article run takes about 5 seconds.
• Real article URLs, not redirects. Google News RSS only gives news.google.com/rss/articles/... links. The actor exchanges each one for the publisher URL, the same way the Google News website does when you click a story. In testing, every link resolved (3,000 of 3,000 in the largest cloud run).
• Every search operator. "exact phrase", OR, -exclude, site:reuters.com, intitle:, inurl:, when:7d, after: / before: all work, because the query goes to Google News unchanged.
• More than 100 results per query. Google News stops at about 100 articles per search. Turn on Split searches by day and a 7-day query runs once per day: one test returned 575 unique OpenAI articles from a single query.
• Many countries in one run. Add editions such as US:en, GB:en, DE:de, JP:ja and every query, topic and location runs for each.
• Built for monitoring. Duplicates across queries, topics and editions are removed. Only new articles remembers what earlier runs returned, so a scheduled run sends you each article once.
• Polite to publishers. Publisher pages are read one at a time per site, at least a second apart, and robots.txt is honoured by default.
• You pay for results, not attempts. RUN_SUMMARY lists what happened to every feed, and failures cost nothing.

Use cases

• PR and comms teams: track brand, CEO and competitor mentions across countries, with daily only-new alerts to Slack or email.
• Media monitoring and clipping: collect every article on a topic with the publisher URL, author and image for your reports.
• Finance and trading: pull headlines for tickers and companies ("Nvidia" OR NVDA when:1h) into a model or dashboard.
• AI agents and RAG: feed current news, with full text, into an LLM pipeline. Works from Make, Zapier, n8n, LangChain and Apify's MCP server.
• Research and data science: build news datasets by date range, topic or country.
• SEO and content teams: see who ranks in Google News for your keywords.

How to use it

1. Add Search queries, one per line, and/or pick Topics, add Local news locations or paste Google News URLs.
2. Set the Editions (country:language, US:en by default) and a Time window or date range.
3. Choose limits, and turn on Fetch article details or Include full article text if you need them.
4. Click Start, then download the dataset as JSON, CSV, Excel or XML, or read it with the Apify API.

Input example: two searches from the past 24 hours

{
  "queries": ["artificial intelligence", "\"electric vehicles\""],
  "timeWindow": "1d",
  "maxItemsPerQuery": 25
}

Input example: brand monitoring in three countries, only new articles, with details

{
  "queries": ["\"Acme Corp\" OR \"Acme Inc\"", "\"Jane Doe\" Acme"],
  "editions": ["US:en", "GB:en", "IN:en"],
  "timeWindow": "1d",
  "onlyNew": true,
  "stateKey": "acme-brand",
  "fetchArticleDetails": true
}

Schedule it daily or hourly with Apify Schedules. Each run returns only articles the earlier runs didn't.

Input example: Technology and Business headlines for the US and Germany, with full text

{
  "topics": ["TECHNOLOGY", "BUSINESS"],
  "editions": ["US:en", "DE:de"],
  "maxItemsPerQuery": 20,
  "includeArticleText": true
}

Input example: every Reuters article about Tesla in October 2026, beyond the 100 cap

{
  "queries": ["tesla site:reuters.com"],
  "dateFrom": "2026-10-01",
  "dateTo": "2026-11-01",
  "splitByDay": true,
  "maxItemsPerQuery": 0
}

Input fields

Field	What it does
queries	Google News searches, one per line. Operators: "phrase", OR, -word, site:, intitle:, inurl:, when:, after:, before:
topics	TOP_STORIES, WORLD, NATION, BUSINESS, TECHNOLOGY, ENTERTAINMENT, SPORTS, SCIENCE, HEALTH
locations	Local headlines for a place name: London, Austin, Texas, Bavaria
googleNewsUrls	Any news.google.com search, topic or publication URL (or its RSS URL), or a topic ID. A URL with its own hl/gl/ceid keeps that edition
editions	COUNTRY:language (Google's ceid), e.g. US:en, GB:en, IN:en, CA:fr, DE:de, FR:fr, ES:es, BR:pt-419, MX:es-419, JP:ja, KR:ko, TW:zh-Hant. hl and gl are derived
language	Optional hl override for every edition
timeWindow	1h, 12h, 1d, 2d, 7d, 30d, 1y, or empty for any time. Sent as when: for searches, applied by publish date for other feeds
dateFrom, dateTo	YYYY-MM-DD, sent as after: / before:. Override the time window
splitByDay	Run each search once per day of the window or date range, to get past Google's ~100-per-search cap
maxItemsPerQuery	Max new articles per feed (per edition, and per day when split). 0 = all. Default 100
maxItems	Max articles in the whole run. 0 = no limit. Default 1,000
resolveUrls	Decode Google links to publisher URLs (default on)
fetchArticleDetails	Read each publisher page for description, image, author, date, site name, language
includeArticleText	Also extract the main text and word count (turns on details)
respectRobotsTxt	Skip publisher pages that robots.txt disallows (default on)
dedupe	Return an article once even if several feeds find it (default on)
onlyNew, stateKey	Skip articles returned by earlier runs with the same key (kept 45 days in the key-value store google-news-state)
proxyConfiguration, useProxyForPublishers	Optional Apify Proxy. Google News RSS normally works without one

Output

{
  "title": "iPhone Duo vs. Galaxy Z Fold 8 Camera Comparison: Frolicking With Apple's and Samsung's Foldables",
  "publisher": "CNET",
  "publishedAt": "2026-11-07T13:00:09Z",
  "articleUrl": "https://www.cnet.com/tech/mobile/iphone-duo-vs-galaxy-z-fold-8-camera-comparison/",
  "snippet": "I snapped some photos in a dreamy garden with these passport-size phones to see how their rear cameras stack up.",
  "googleNewsUrl": "https://news.google.com/rss/articles/CBMiiAFBVV95cUxQZVNjbEJN...?oc=5",
  "publisherUrl": "https://www.cnet.com",
  "type": "article",
  "feedType": "topic",
  "query": null,
  "searchQuery": null,
  "topic": "TECHNOLOGY",
  "location": null,
  "day": null,
  "edition": "US:en",
  "hl": "en-US",
  "gl": "US",
  "feedUrl": "https://news.google.com/rss/headlines/section/topic/TECHNOLOGY?hl=en-US&gl=US&ceid=US%3Aen",
  "rank": 1,
  "articleId": "CBMiiAFBVV95cUxQZVNjbEJN...",
  "relatedArticles": [
    {"title": "The iPhone Duo has more than 100,000 optimized apps, Apple says", "publisher": "The Verge", "googleNewsUrl": "https://news.google.com/rss/articles/..."}
  ],
  "detailsStatus": "ok",
  "articleTitle": "iPhone Duo vs. Galaxy Z Fold 8 Camera Comparison: Frolicking With Apple's and Samsung's Foldables - CNET",
  "description": "I snapped some photos in a dreamy garden with these passport-size phones to see how their rear cameras stack up.",
  "imageUrl": "https://www.cnet.com/wp-content/uploads/sites/2/iPhone-Duo-vs.-Galaxy-Z-Fold-8-camera-comparison.jpg",
  "author": "Abrar Al-Heeti; CNET",
  "articlePublishedAt": "2026-11-07T08:00:09-05:00",
  "siteName": "CNET",
  "language": "en-US",
  "keywords": ["apple", "samsung"],
  "text": "There are a handful of things Apple's iPhone Duo and Samsung's Galaxy Z Fold 8 have in common...",
  "wordCount": 875,
  "scrapedAt": "2026-11-08T03:42:27Z"
}

Field	Notes
title	Headline with Google's trailing - Publisher removed
publishedAt	ISO 8601 in UTC, from the feed. Older articles sometimes carry a date only (midnight US Pacific)
articleUrl	The publisher's article URL. null if it could not be decoded (that article is free) or if resolveUrls is off
snippet	Google's RSS has no snippet text. With article details on, this is the publisher's own description
query / searchQuery	Your query, and the exact q sent to Google with time operators added
day	The day slice, when Split searches by day is on
rank	Position in the Google News feed
relatedArticles	Topic and top-stories feeds group coverage of one story; the other outlets are listed here
detailsStatus	ok, or why the page was not read: error: HTTP 403, skipped: disallowed by robots.txt... Only ok pages are charged
Details fields	articleTitle, description, imageUrl, author, articlePublishedAt, siteName, language, keywords, plus text and wordCount with full text on. They appear only when details are on

The dataset has three views: Articles, Article details and Monitoring (compact). The key-value store record RUN_SUMMARY lists each feed's status, how many articles it had, duplicates removed, links resolved, pages read and any errors.

Pricing

Pay per event. You pay only for what comes back:

Event	Price	When
Article	$0.002	Once per article row in the dataset. The decoded article URL is included. Articles whose link could not be decoded are free
Article details	$0.001	Once per article whose publisher page was read successfully (details and full text cost the same). Blocked, paywalled, robots-disallowed and failed pages are free

There is no start fee. A feed that fails costs nothing. An empty input returns up to 50 Top stories for the US as a sample, so at most $0.10. Set Max cost per run in the run options to cap spending: the actor stops cleanly before an article that would go over the limit.

Examples:

• 10 brand queries, 25 new articles each, every day for a month: 7,500 articles = $15
• One-off pull of 1,000 articles with full text: about $3
• Hourly headlines for 3 topics in 2 countries, only new (about 300 new articles a day): about $18 a month

Tips

• Put the time filter in the window, not the query, unless you need something unusual: a query with its own when:, after: or before: keeps it and ignores the window.
• Exact phrases cut noise. "interest rates" finds far fewer off-topic articles than interest rates.
• Editions change the results. GB:en and US:en give different publishers and rankings for the same query. Add each market you care about.
• Use site: for one publisher, e.g. tesla site:reuters.com. Note that some publishers only appear in some editions.
• Use a separate stateKey per monitor, so two schedules don't hide each other's articles.
• Full text is best-effort. Paywalled sites (WSJ, FT, NYT...) usually return 401/403 or just the intro. Those pages are free and marked in detailsStatus.

Other actors from Tidewell

• ATS Jobs API: jobs from Greenhouse, Lever, Ashby, Workable and 4 more ATSes
• Document to Markdown for LLM & RAG: PDF, Word, PowerPoint, Excel and scans to Markdown or RAG chunks
• Google Trends API: interest over time, rising queries and Trending Now
• Website Screenshot & PDF API: full-page screenshots and URL to PDF in real Chrome

FAQ

Is scraping Google News legal? What about Google's terms? The actor reads Google's public RSS feeds and the public article pages they link to, and collects no personal data beyond what publishers print in a byline. Google's feed terms say the feeds are for personal, non-commercial use in a feed reader, and Google's Terms of Service restrict automated access. Using this actor may therefore conflict with Google's terms, and article text belongs to its publishers. You are responsible for how you use the data, including copyright in article text: storing headlines and links for monitoring is a very different thing from republishing articles. If you need a licensed feed, use a commercial news API. This is not legal advice.

Why are some article URLs null? Google occasionally won't decode a link (rare: every link in our tests resolved). The article is still returned with its Google News link, and you are not charged for it.

Why do I get fewer than 100 results? Google News returns at most about 100 articles per search, fewer for narrow queries or short windows. Topic and location feeds return up to about 70. Use Split searches by day or more editions to get more.

Can I get articles older than a month? Yes. Use dateFrom / dateTo (or after: / before: in the query). Coverage of older dates is thinner than for recent news, and the times of older articles are often date-only.

How fresh is the data? The feeds are live: a when:1h search returns articles from the past hour. Hourly schedules with Only new articles work well for alerts.

What happens when Google rate-limits? The actor backs off and retries. With Apify Proxy on, it also moves to a new IP. A feed that still fails is reported in RUN_SUMMARY, not charged, and the run continues. If every feed fails, the run fails with a clear message, so a schedule doesn't silently store nothing.

Something broke or looks wrong? Open an issue on the actor's Issues tab with the run ID. When Google changes something, the fix ships here.

Keywords

google news api, google news scraper, google news rss, news scraper, news api, news monitoring, media monitoring, brand monitoring, headlines api, google news search, google news topics, decode google news url, google news redirect, real article url, article text extraction, full text news, news alerts, press monitoring, pr monitoring, n8n news, ai news agent