News Search: Retrieve Recent News Articles via API

The Custom Search JSON API does not expose a dedicated news tab, but you can reliably surface timely journalism by combining sort=date (most-recently-indexed first), dateRestrict (a rolling time window), and siteSearch (targeting specific news outlets or Google News–indexed domains). This approach mirrors the behavior of a news search tab and gives you precise control over which sources, time ranges, and topics your application surfaces.

How news search works

When you set sort=date, Google returns results ordered by the date each page was indexed rather than by relevance score. Pairing that with dateRestrict ensures you only retrieve articles published within a defined recency window. You can further sharpen results by restricting to known news domains (e.g., reuters.com, bbc.com, apnews.com) via siteSearch, or by configuring your Programmable Search Engine in the Control Panel to include a curated list of news outlets. Publication dates are not returned as a dedicated API field. Instead, they appear in the pagemap object — typically in metatags[0]["article:published_time"] (ISO 8601) or in metatags[0]["og:updated_time"] — and sometimes in the human-readable snippet text.

Google News (news.google.com) content itself is not accessible through the Custom Search API. Use siteSearch to target the original publisher domains where the articles are hosted.

Key parameters for news search

string

required

Your news query. Standard Google operators work — for example:

q=inflation+rate — broad keyword match
q="interest+rate"+decision+intitle:central+bank — exact phrase plus title match
q=earthquake+OR+tsunami — either term

string

Set to date to order results by index date (most recent first) instead of relevance. This is the single most important parameter for news retrieval.

string

Limit results to content indexed within a rolling window. Format: d[N], w[N], m[N], y[N].

Value	Window
`d1`	Last 24 hours
`d7`	Last 7 days
`w2`	Last 2 weeks
`m1`	Last month
`m3`	Last 3 months
`y1`	Last year

For breaking news, use d1 or d2. For trend analysis, use m1 or m3.

string

Restrict results to a specific news outlet domain. Combine with siteSearchFilter=i to include only that source. You can only specify one domain per request; to search across multiple outlets, configure them as included sites in your engine.

string

i — include only the siteSearch domain. e — exclude that domain. Defaults to i.

string

Require this exact phrase to appear in every returned result. Useful for filtering out loosely related articles when your query terms are ambiguous.

string

Exclude any result that contains this term. Useful for filtering out topic crossover (e.g., excluding sports from results for "league table").

string

Restrict results to a specific language. Format: lang_XX (e.g., lang_en, lang_de, lang_fr). See Advanced Filters for the full list.

string

Geolocation of the end user. Two-letter country code (e.g., us, gb, in). Influences which regional editions of news sites rank higher.

integer

Results per page. Accepts 1–10. Defaults to 10.

integer

Pagination offset (1-based, max 91).

Make your first news search request

Pick your recency window

Decide how fresh the content needs to be. For a real-time news widget, use d1; for a weekly digest feature, use w1 or d7.

Choose a source strategy

Decide between:

Single outlet — siteSearch=reuters.com&siteSearchFilter=i
Multi-outlet engine — Configure trusted domains in the Programmable Search Engine Control Panel, then omit siteSearch
Open web news — Omit siteSearch entirely; add sort=date and dateRestrict to surface any recently indexed news

Send the request

curl "https://www.googleapis.com/customsearch/v1\
?key=YOUR_API_KEY\
&cx=YOUR_CX\
&q=federal+reserve+interest+rate\
&sort=date\
&dateRestrict=d7\
&siteSearch=reuters.com\
&siteSearchFilter=i\
&num=10"

Extract the publication date

Parse the pagemap.metatags[0]["article:published_time"] field from each result item. Fall back to the snippet text if the structured field is absent.

Full curl examples

Breaking news (last 24 h)
Single outlet, topic search
Technology news, last month
Exclude specific terms

Retrieve the most recent articles about an earthquake from any indexed source:

curl "https://www.googleapis.com/customsearch/v1\
?key=YOUR_API_KEY\
&cx=YOUR_CX\
&q=earthquake\
&sort=date\
&dateRestrict=d1\
&lr=lang_en\
&num=10"

Fetch the last week of AP News articles on US elections:

curl "https://www.googleapis.com/customsearch/v1\
?key=YOUR_API_KEY\
&cx=YOUR_CX\
&q=US+election\
&sort=date\
&dateRestrict=w1\
&siteSearch=apnews.com\
&siteSearchFilter=i\
&num=10"

Surface recent technology news from multiple outlets (engine pre-configured):

curl "https://www.googleapis.com/customsearch/v1\
?key=YOUR_API_KEY\
&cx=YOUR_CX\
&q=artificial+intelligence+regulation\
&sort=date\
&dateRestrict=m1\
&lr=lang_en\
&gl=us\
&num=10"

Search for climate news but exclude sports-related crossover results:

curl "https://www.googleapis.com/customsearch/v1\
?key=YOUR_API_KEY\
&cx=YOUR_CX\
&q=climate+change\
&sort=date\
&dateRestrict=m1\
&excludeTerms=sports\
&num=10"

Sample news result item

{
  "kind": "customsearch#result",
  "title": "Federal Reserve Holds Rates Steady Amid Inflation Concerns",
  "htmlTitle": "<b>Federal Reserve</b> Holds Rates Steady Amid <b>Inflation</b> Concerns",
  "link": "https://www.reuters.com/markets/us/fed-holds-rates-steady-2024-11-08/",
  "displayLink": "www.reuters.com",
  "snippet": "Nov 8, 2024 — The Federal Reserve left its benchmark interest rate unchanged on Wednesday, citing continued uncertainty about the inflation outlook as policymakers weigh when to begin cutting...",
  "htmlSnippet": "Nov 8, 2024 — The <b>Federal Reserve</b> left its benchmark interest rate unchanged on Wednesday...",
  "formattedUrl": "https://www.reuters.com/markets/us/fed-holds-rates-steady-2024-11-08/",
  "pagemap": {
    "metatags": [
      {
        "og:title": "Federal Reserve Holds Rates Steady Amid Inflation Concerns",
        "og:description": "The Federal Reserve left its benchmark interest rate unchanged on Wednesday...",
        "og:url": "https://www.reuters.com/markets/us/fed-holds-rates-steady-2024-11-08/",
        "og:image": "https://www.reuters.com/resizer/fed-image.jpg",
        "article:published_time": "2024-11-08T18:04:00Z",
        "article:modified_time": "2024-11-08T19:45:00Z",
        "article:author": "Reuters Staff",
        "article:section": "Markets"
      }
    ],
    "newsarticle": [
      {
        "datePublished": "2024-11-08T18:04:00Z",
        "headline": "Federal Reserve Holds Rates Steady Amid Inflation Concerns",
        "description": "The Federal Reserve left its benchmark interest rate unchanged on Wednesday."
      }
    ]
  }
}

How to read publication dates

Publication dates are not a dedicated top-level API field. Extract them using this priority order:

1. article:published_time (most reliable)

Available in pagemap.metatags[0]["article:published_time"] when the publisher uses the Open Graph Article protocol. Returns an ISO 8601 timestamp:

const publishedAt = item.pagemap?.metatags?.[0]?.["article:published_time"];
// e.g. "2024-11-08T18:04:00Z"

2. newsarticle.datePublished

Some publishers include Schema.org NewsArticle markup. Check pagemap.newsarticle[0].datePublished:

const publishedAt =
  item.pagemap?.newsarticle?.[0]?.datePublished ??
  item.pagemap?.metatags?.[0]?.["article:published_time"];

3. og:updated_time

Falls back to pagemap.metatags[0]["og:updated_time"] when article:published_time is absent:

const updatedAt = item.pagemap?.metatags?.[0]?.["og:updated_time"];

4. Snippet date prefix

Many news snippets begin with the publication date in the format "Nov 8, 2024 — ...". Parse it as a last resort:

const dateMatch = item.snippet?.match(/^([A-Z][a-z]+ \d{1,2}, \d{4})/);
const fallbackDate = dateMatch ? new Date(dateMatch[1]) : null;

pagemap data reflects what Google extracted at crawl time. A very recently published article may not yet have its pagemap fully populated. Use the snippet date prefix as a fallback for breaking-news use cases where freshness is critical.

Build a news monitoring workflow

Set a polling interval

For near-real-time monitoring, poll the API every 15–30 minutes using dateRestrict=d1 and sort=date. This keeps you within the free tier of 100 queries/day if you poll every ~15 minutes.

Track seen articles

Store the link (canonical URL) of each returned article in a database or cache. On each poll, skip any item whose link you have already processed.

Filter by publication date

After extracting article:published_time, discard any articles published before your last poll timestamp to prevent duplicates during recrawl windows.

Handle pagination for broad queries

If a topic is fast-moving, the first page may fill quickly. Page through results (start=11, start=21, …) until you encounter an article you’ve already seen.

For high-volume news monitoring on the paid tier (up to 10,000 queries/day per project), use a caching layer in front of the API. Cache results by q + dateRestrict + siteSearch for 5–10 minutes to avoid duplicate calls for the same topic from concurrent users.

Error reference

HTTP status	Meaning	Common cause
`400`	Bad request	Malformed `dateRestrict` value or missing `q`
`401`	Unauthorized	Invalid or missing API key
`403`	Forbidden	Daily quota exceeded
`429`	Too many requests	Burst rate limit; back off and retry
`500`	Server error	Transient issue; retry with exponential back-off

Get Started

Core Concepts

Search Features

Guides

Help

News Search: Retrieve Recent News Articles via API

How news search works

Key parameters for news search

Make your first news search request

Full curl examples

Sample news result item

How to read publication dates

Build a news monitoring workflow

Error reference

​How news search works

​Key parameters for news search

​Make your first news search request

​Full curl examples

​Sample news result item

​How to read publication dates

​Build a news monitoring workflow

​Error reference

How news search works

Key parameters for news search

Make your first news search request

Full curl examples

Sample news result item

How to read publication dates

Build a news monitoring workflow

Error reference