NewsGuard: since May 2023, websites hosting AI-created fake news articles have grown from 49 to 600+; some post “hundreds if not thousands of articles” per day
Pranshu Verma / Washington Post : Mastodon: @jeffjarvis@mastodon.social . X: @pranshuverma_ . LinkedIn: Franky Saegerman Mastodon: Jeff Jarvis / @jeffjarvis@mastodon.social : The issue isn't volume of disinformation; there's already more than enough. Nor quality; people say they believe stupid shit. The problem is their desire to belong to the cult.... The rise of AI fake news is creating a ‘misinformation superspreader’ https://www.washingtonpost.com/ ... X: Pranshu Verma / @pranshuverma_ : new: Since May, AI fake news sites have ballooned by over 1,000%. They can churn out misinfo about elections, wars + natural disasters far quicker than troll farms & propaganda networks. “This is why we call it the next great misinformation superspreader."https://www.washingtonpo st.com/ ... LinkedIn: Franky Saegerman : As I mentioned at the #GSMCON2023 conference earlier this year , AI is making it easy for anyone to create propaganda outlets, producing content that can be hard to differentiate from real news. …
Context & Ripple Effects
NewsGuard's May 2023 baseline identified 49 generative-AI sites presenting themselves as news outlets, some carrying falsehoods and programmatic ads. This report turns that early signal into a rapid scaling pattern: the initial 49-site finding has become a much larger publishing network.
The expansion fits prior evidence that ChatGPT could produce polished text that repeated conspiracy theories and misleading narratives. Convincing low-cost text generation lowers the labor needed to sustain sites that resemble conventional publishers.
First-order effects
- NewsGuard and other trust-and-safety teams face a far larger set of domains and articles to identify, assess, and track, particularly where sites publish at industrial volume.
- Readers, advertisers, and distribution services encounter more outlets whose news-like presentation can mask unreliable or false material.
Second-order effects
- Search, social, and ad platforms face greater pressure to distinguish bulk-generated publisher content from legitimate reporting, because site-level review becomes less sufficient as output volume rises.
- The economics of programmatic publishing become more attractive to operators able to generate and distribute articles cheaply, intensifying competition for attention and ad inventory.
Third-order effects
- If this pattern persists, online news credibility may depend less on the apparent polish or frequency of publication and more on provenance, verification, and durable publisher reputation.
- The development is an example of a synthetic-supply problem: automated content can expand faster than the institutions and systems built to evaluate it, potentially shifting platform policy toward stronger source-level controls.
The trend: Generative AI is turning misinformation from a labor-intensive campaign activity into a scalable content-supply and distribution challenge.