HOW WE FIND, VERIFY AND TRACK ARTIFICIAL INTELLIGENCE DEVELOPMENTS
The sourcing, inclusion, classification and correction standards behind the Artificial Intelligence Tracker.
The Artificial Intelligence Tracker is not intended to collect every AI announcement. It records developments that materially change the trajectory of artificial intelligence—or reveal where that trajectory is going.
THE INCLUSION STANDARD
A development qualifies when credible evidence indicates that it materially changes or reveals one or more of the following:
- AI capabilities or limitations
- Adoption, productivity or employment
- Economics, investment or infrastructure
- Competition among companies, models or nations
- Safety, control, cybersecurity or rogue behavior
- National power, warfare or social institutions
Routine product updates, promotional claims and repetitive commentary are excluded unless they provide meaningful new evidence.
HOW WE VERIFY DEVELOPMENTS
The Institute gives priority to primary evidence: company announcements, technical reports, research papers, regulatory filings, government documents, official data and direct statements from the people or organizations involved.
Credible independent reporting is used when primary material is unavailable, incomplete or requires additional context. Consequential claims are checked against more than one reliable source whenever possible.
HOW WE CLASSIFY THE EVIDENCE
Each development is separated into three layers:
- What happened — the verifiable event or new evidence.
- What it means — the immediate implications supported by that evidence.
- What may come next — the Institute’s assessment of the larger pattern and possible consequences.
Facts, reported claims and Institute analysis are distinguished from one another. Uncertainty is stated rather than concealed.
UPDATES AND CORRECTIONS
Entries may be expanded, reclassified or corrected as better evidence becomes available. A tracker is a living intelligence record—not a frozen archive.