Google News Monitor: New Articles and AI Tags

Track companies, people, keywords and topics on Google News, in any country and language. Each article is one row with the headline, publisher, publish time, the queries it matched and a Google News link.

Turn on monitor mode and schedule the Actor hourly or daily: each run returns only the articles that earlier runs did not return, so you see (and pay for) new articles only. Add AI tags to label each article as funding, layoffs, acquisition, leadership change, product launch and more, with sentiment and the companies named, and keep only the event types you care about.

For AI agents

Use this Actor when you need recent news articles about named companies, people, products or keywords, or the top headlines of a Google News section, in a given country and language. Not for full article text. Rows have the headline, publisher, date and link, not the article body.

• Actor id: brightknot/google-news-monitor
• Required input: none (give at least one of the main list fields)
• Call: MCP call-actor with the input below, then get-dataset-items. Over HTTP: POST https://api.apify.com/v2/acts/brightknot~google-news-monitor/run-sync-get-dataset-items with the input as the JSON body.
• Cap the spend with maxTotalChargeUsd in the call options, not in the input.

Example input

{
  "queries": ["\"Stripe\"", "\"Revolut\" funding OR layoffs"],
  "editions": ["US:en", "GB:en"],
  "when": "7d",
  "maxArticlesPerQuery": 20,
  "monitorMode": false
}

Example output row

Trimmed to the main fields. See the Output schema tab for every field.

{
  "article_id": "0ff8cb2d97376dda4c3c",
  "query": "\"Stripe\"",
  "edition": "US:en",
  "title": "Stripe raises $1.2 billion at a $110 billion valuation",
  "source_name": "TechCrunch",
  "source_domain": "techcrunch.com",
  "published_at": "2026-10-13T07:56:46Z",
  "google_news_url": "https://news.google.com/rss/articles/CBMiTGh0dHBzOi8vdGVjaGNydW5jaC5jb20vMjAyNi8xMC8xMS9zdHJpcGUtcmFpc2VzLTEtMi1iaWxsaW9uL9IBAA?oc=5",
  "related_coverage_count": 2,
  "matched_queries": ["\"Stripe\""],
  "editions": ["US:en"]
}

Rows are newest first within each query and edition. A SUMMARY record shows the status of every feed.

Cost per call

Event	Price	When
Article	USD 0.003	per article row returned
AI-tagged article	USD 0.002	per article labelled, only with aiTags on
Actor start	Apify default	once per run

The example input returns at most 80 rows (2 queries x 2 editions x 20), so it costs at most USD 0.24.

Limits

• About 100 articles per query per edition at most (Google News feed limit).
• Usually finishes in under a minute for a few queries; AI tags add about 10 to 30 seconds.
• monitorMode defaults to true. Set it to false for one-off searches, or a repeat call can return 0 rows.
• No credentials needed. Runs with limited permissions.

What it does

• Searches Google News for each query, with full Google News operators: exact names in quotes, OR, -word, site:, intitle:.
• Reads top headlines for sections such as Business or Technology.
• Reads one or many editions, such as US:en, GB:en, IN:en, AE:ar or DE:de.
• Returns each article once, even when it matches several queries, topics or editions, and lists every match.
• Filters by publisher (include or exclude) and by exact article age.
• Optional AI tags: event type, sentiment and companies named, from a small AI model.
• Saves a run summary with counts per feed, failed feeds, duplicates merged and event type counts.

Use cases

• Sales teams: watch target accounts and get only new funding, expansion and leadership change news, the moments to reach out.
• PR and comms: track your brand and competitors in several countries every hour, with negative stories flagged.
• Investors and analysts: a daily feed of acquisitions, layoffs and earnings news for a watchlist of companies.
• Market research: every article about a topic or product category in the last 30 days, in one table.

Input

Field	What it does
Search queries	One per line. A company, keyword or phrase, with Google News operators if you like: "Stripe", "Revolut" funding OR layoffs, site:reuters.com fintech.
Topic headlines	Top stories for World, Nation, Business, Technology, Entertainment, Sports, Science or Health.
Editions	COUNTRY:language, e.g. US:en, GB:en, AE:ar, DE:de. Default US:en.
Time window	Last hour, 12 hours, day, 7 days (default), 30 days, year or any time. Applies to search queries.
Only articles from the last N hours	Exact age filter for queries and topics.
Max articles per query per edition	Up to 100, newest first.
Only these publishers / Skip these publishers	Match on publisher name or domain, e.g. reuters, prnewswire.
One row per article	Merge duplicates across queries, topics and editions. On by default.
Monitor mode	Only return articles not returned by earlier runs with the same monitor name. On by default.
Monitor name	A separate memory of seen articles for each name.
AI tags	Adds ai_event_type, ai_sentiment and ai_companies.
Only these event types	Keep only these AI event types, e.g. funding and acquisition. Turns AI tags on.
AI model	OpenRouter model id for AI tags. Default anthropic/claude-haiku-5.5.

Example input:

{
  "queries": ["\"Stripe\"", "\"Revolut\"", "\"Adyen\""],
  "editions": ["US:en", "GB:en"],
  "when": "1d",
  "excludeSources": ["prnewswire", "globenewswire"],
  "monitorMode": true,
  "monitorName": "fintech-accounts",
  "eventTypes": ["funding", "acquisition", "leadership_change", "layoffs"]
}

Output (one row per article)

{
  "change_type": "new",
  "article_id": "4f1c2a9d0b7e6a5c3d21",
  "query": "\"Stripe\"",
  "topic": null,
  "edition": "US:en",
  "title": "Stripe raises $1.2 billion at a $110 billion valuation",
  "source_name": "TechCrunch",
  "source_domain": "techcrunch.com",
  "source_url": "https://techcrunch.com",
  "published_at": "2026-10-12T05:00:00Z",
  "age_hours": 2.0,
  "google_news_url": "https://news.google.com/rss/articles/CBMiTGh0dHBz...?oc=5",
  "related_coverage_count": 2,
  "matched_queries": ["\"Stripe\""],
  "editions": ["US:en", "GB:en"],
  "ai_event_type": "funding",
  "ai_sentiment": "positive",
  "ai_companies": ["Stripe"],
  "scraped_at": "2026-10-12T07:00:00Z"
}

Field	Meaning
change_type	new in monitor mode. Empty when monitor mode is off.
query / topic	The first query or topic that found the article. All matches are in matched_queries (topics as topic:BUSINESS).
edition, editions	The first edition that returned the article, and all editions that did.
title	The headline without the " - Publisher" ending Google adds.
source_name, source_domain, source_url	The publisher as Google News lists it.
published_at, age_hours	Publish time in UTC and its age at run time.
google_news_url	The Google News link. It opens the article on the publisher's site.
related_coverage_count	How many other articles Google groups with this story. A rough measure of how big the story is.
ai_event_type	funding, acquisition, layoffs, hiring_expansion, leadership_change, product_launch, partnership, earnings, legal_regulatory, security_incident or other. Only with AI tags.
ai_sentiment	positive, neutral or negative for the main company in the headline. Only with AI tags.

A run summary is saved as SUMMARY in the run's key-value store: articles read, returned, filtered out, already seen and merged, the status and article count of every feed, and event type counts.

Pricing

$3.00 per 1,000 articles, plus $2.00 per 1,000 AI-tagged articles when AI tags are on.

You pay for each article row in the results. Filters run before output, so filtered articles are not charged. In monitor mode a scheduled run only pays for articles that are new since the last run. AI tags are charged for each article the model labels, and the small model cost (about $0.0005 per 20 headlines with Claude Haiku 5.5, an estimate) is billed by Apify's OpenRouter proxy. Example: 20 companies checked daily with about 5 new articles each is 100 articles a day, about $0.30 a day, or $0.50 with AI tags.

How it works

1. Builds one Google News RSS feed per query (or topic) and edition, adding your time window to each query.
2. Reads the feeds a few at a time with a short pause, and retries rate limit and server errors with backoff.
3. Parses each article, removes the publisher suffix from the headline, converts dates to UTC and applies your filters.
4. Merges the same article across feeds, using Google's article id.
5. In monitor mode, skips articles saved by earlier runs under your monitor name (named key-value store google-news-monitor-state), then saves the new ones.
6. With AI tags, sends the new headlines in batches of 20 to a small model through Apify's OpenRouter proxy.

Google News RSS feeds are public. No login, no cookies and no browser are used.

Limits

• Google News returns at most about 100 articles per feed. For busy topics, use narrower queries, a shorter time window or several editions, and schedule runs more often.
• Article links are Google News links that open the publisher's page. The article text is not downloaded.
• AI tags are based on the headline and publisher only, so they can be wrong for vague headlines.
• Google can rate limit heavy use. Keep the number of feeds per run reasonable (queries x editions) and schedule rather than run everything at once.

FAQ

How do I search for an exact company name? Put it in double quotes: "Acme Robotics". Without quotes Google also matches the words separately.

Can I monitor news in Arabic or German? Yes. Use editions such as AE:ar, SA:ar, DE:de or AT:de.

Does the first monitor run return everything? Yes, everything in the time window up to your maximum. Later runs return only new articles.

Why did one article match several queries? With "One row per article" on, it is returned once and matched_queries lists every query that found it.

Do I need an AI account for AI tags? No. The model cost is billed through your Apify account by the OpenRouter proxy.