How we verify AI claims
AI Ledger reviews trending AI-industry claims by tracing each one back to primary sources — company announcements, court filings, and named reporting — rather than aggregating secondhand summaries. Every claim gets one of four verdicts: Verified, Disputed, Unverified, or Debunked, published with the evidence for and against, and corrected in place when new information changes the picture.
Verified
Confirmed by a primary source (the company or institution itself) or corroborated by multiple independent, named outlets with no credible contradiction.
Disputed
Credible sources disagree, or the claim is confirmed in part but contested in a material detail (scale, timing, attribution).
Unverified
No primary source has confirmed or denied it — including forward-looking predictions and forecasts, which are opinions about the future, not settled facts, no matter how credible the source.
Debunked
Directly contradicted by a primary source or verifiable record. We explain what's actually true, not just what's false.
What we don't do
We don't adjudicate opinions, predictions, or matters of taste as if they were facts — a forecast stays "Unverified" until it either happens or is confirmed by the party making it, no matter how many analysts repeat it. We review one atomic claim per page; we don't publish general commentary or editorializing about the AI industry.
Corrections
Every claim page shows a "reviewed by" line with the last update date. When new evidence changes a verdict, we edit the page in place, bump the date, and note what changed — we do not silently delete or rewrite history. Spot an error? Tell us and we'll re-review it.
Editorial independence
AI Ledger is not owned by, or affiliated with, any AI lab we review. Ad revenue funds this site, but verdicts are never influenced by advertisers — a company can be a claim's subject and never a factor in its rating.