AI & Media
How many AI-generated news articles are published every day?
A live count of AI-generated and AI-assisted news articles published globally
Roughly 41.7 articles every minute.
Source: NewsGuard AI News Tracker 2024; AP, Bloomberg usage data. View on dashboard →
What is AI-generated news?
AI-generated news means articles, summaries, and reports that are mostly or entirely written by large language models, with little human editing. This took off in 2022 when LLMs became widely available. Reuters Institute surveys show rising public concern. AP and Bloomberg have used automated article generation since 2014 for earnings and sports. The 2023-2024 shift: AI can now produce opinion, analysis, and long-form journalism at scale. Misinformation farms run hundreds of fake news sites, each publishing dozens of AI-written articles daily.
What this means for you
NewsGuard, the news reliability tracker, identified over 1,265 AI-generated content farms by 2024: websites that produce hundreds of articles per day with no human editorial oversight. These sites rank in search results, appear in news feeds, and are cited as sources. You have almost certainly read content from one without knowing it.
The challenge is not that AI-assisted journalism is inherently low-quality. The AP, Bloomberg, and Reuters all use AI tools to produce financial reports and sports recaps, with editorial standards intact. The problem is the content farm model: AI used specifically to produce high volumes of loosely-accurate content designed to capture search traffic, not to inform.
A practical check: look for author names, publication dates, source citations, and editorial standards disclosures. Sites that lack all of these, especially those covering multiple unrelated topics, are statistically likely to be AI-generated content farms. Your media literacy is now a direct defence against this kind of content.
AI journalism: from automated earnings reports to misinformation factories
From automation to journalism
Algorithmic journalism has a longer history than most assume. AP's Automated Insights partnership began generating quarterly earnings reports in 2014, producing 3,000 articles per quarter without human writers. These were accurate, efficient, and indistinguishable in quality from manually-written financial summaries. Bloomberg uses its Cyborg system for financial news. The 2022-2023 LLM revolution expanded this from structured data to unstructured journalism, and simultaneously enabled bad actors to operate content farms at unprecedented scale.
The content farm threat
NewsGuard documented over 1,000 websites in 2023 that look like AI content farms: dozens to hundreds of articles per day on any topic, built to capture search traffic, not inform. They mix real topics with misinformation, outpace fact-checkers, and earn from ads. The scale is a real problem for how people find reliable news.
The scale of AI-generated content
The Associated Press has used AI to generate earnings report articles since 2014, pioneering automated journalism
Pangram LabsBy 2023, over 60% of Reuters Institute survey respondents were concerned about using AI in news production
Pangram LabsNewsGuard documented 1,265 AI-generated news websites by mid-2024, each publishing dozens to hundreds of AI-written articles daily
NewsGuardCNET published 77 AI-written finance articles in early 2023, 41 required corrections after publication
Pangram LabsStanford HAI AI Index 2026: AI-generated content concerns are rising globally; most capable frontier models no longer disclose training data, parameter counts, or training code, raising transparency issues
Stanford University Human-Centered AIAI news articles vs. fake news shared, today
As AI coverage grows, so does the spread of misinformation about AI and other topics. AI-generated content farms now produce thousands of articles per day that blur the line.
AI news volume growth trends: 2014-2024
AI news generation started as legitimate automation for financial data in 2014 but transformed into a mass disinformation risk from 2022, when LLMs enabled content farms to produce thousands of convincing but often false articles daily.
| Year | Rate | Est. per day | Context |
|---|---|---|---|
| 2014 | 10/hr | 240 | Structured data automation only |
| 2022 | 500/hr | 12K | CNET, Forbes adopt AI writing |
| 2024 | 3K/hr | 60K | AI content farms scale; 1,265 sites documented by NewsGuard |
| 2027 (forecast) | 6K/hr | 150K | AI content farms proliferate; regulation lags |
How AI news became a media industry issue
- 2014AP partners with Automated Insights to generate earnings report articles, first mainstream automated journalism
- 2016Washington Post deploys Heliograf for election and Olympics coverage
- 2022CNET begins publishing AI-generated finance articles; GPT-4 enables general-purpose journalism AI
- 2023CNET AI journalism scandal (41% correction rate); NewsGuard documents 49 AI content farms
- 20241,000+ AI content farm websites identified; AI-generated content indistinguishable to most readers without disclosure
Research on AI journalism and content automation
| Year | Finding | Value | Source |
|---|---|---|---|
| 2014 | AP launches Wordsmith for earnings report automation; first major mainstream AI journalism deployment | qualitative milestone | Pangram Labs |
| 2022 | Reuters Institute Digital News Report 2022: 62% of journalists surveyed concerned about AI impact on journalism | 62 % journalists concerned about AI | Pangram Labs |
| 2023 | CNET AI journalism scandal; NewsGuard identifies 49 AI content farms; ~60,000 AI articles/day estimated across platforms | 60K AI news articles/day (estimate) | Pangram Labs |
| 2024 | NewsGuard AI Tracking Center: 1,265 unreliable AI-generated news websites documented by mid-2024; model estimate ~60,000 AI news items/day based on site count × estimated median publication rate | 60K AI news articles/day (model estimate) | NewsGuard |
How much AI content is out there: a perspective
One new AI article rolls out every 1.4 seconds. Day and night, all year long.
Printed and stacked, one month of AI news articles would stand taller than the Eiffel Tower.
60,000 AI articles per day is more than Reuters, AP, and AFP combined publish in a week.
In a single year, AI content farms produce more articles than the New York Times has printed in its entire 174-year history.
How the number is calculated
The ~60,000 articles/day figure is a modelled estimate, not a directly measured count. NewsGuard documented 1,265 "unreliable AI-generated news" websites by mid-2024 (AI Tracking Center). Assuming a median of ~47 articles/day per site gives ~59,500 articles/day from content farms alone. Adding AP's documented ~3,700 AI earnings articles/year and other legitimate uses (Bloomberg, Reuters) brings the total to roughly 60,000–80,000 AI-assisted news items daily. The live counter uses 60,000/day ÷ 24 hrs ≈ 2,500/hr. This is a model estimate with significant uncertainty; actual volume could be higher or lower.
Sources: Pangram Labs - AI-Generated Content Research - NewsGuard - AI-Generated News Website Tracker. Methodology →
Frequently asked questions
- How many AI-generated news articles are published daily?
- Estimates vary widely depending on definition. If limited to major outlets using AI assistance (AP, Bloomberg, Reuters), the number is in the thousands daily. If including AI content farms and SEO-driven mass production, estimates of 60,000+ articles per day have been reported.
- Which news organisations use AI to write articles?
- The Associated Press, Bloomberg, Forbes, Reuters, and the Washington Post all use AI-assisted or AI-generated content for certain categories (earnings reports, sports results, weather, election results). AP has used Automated Insights' "Wordsmith" system since 2014.
- Is AI journalism reliable?
- For structured data (sports scores, financial results), AI journalism is accurate and efficient. For general news, bias and hallucination are significant risks. CNET, BuzzFeed, and other outlets that experimented with AI-generated articles reported numerous factual errors when stories were published without adequate review.
How the AI news article estimate is built
The 60,000 articles/day estimate is based on NewsGuard's AI Tracking Center, which systematically monitors AI-generated news websites and documented 1,265 such sites by mid-2024. NewsGuard is an independent media rating organisation whose methodology is publicly documented. The calculation multiplies the documented site count by an estimated median publication rate (not a NewsGuard-published figure), making this a model estimate rather than a directly observed value. AP's automated journalism figures come from publicly disclosed partnership data.
Sources
Pangram Labs - AI-Generated Content Research - NewsGuard - AI-Generated News Website Tracker.
Explore related: Fake news shared - AI hallucinations - Deepfakes created, and the live AnythingCounter dashboard.