OUR EVIDENCE STANDARD.
How AIM26LIVE distinguishes observed data, AI predictions, company statements, peer-reviewed research and independently demonstrated outcomes.
Task exposure is not job displacement.
Our jobs page uses Anthropic's March 2026 observed-exposure measure. The score uses occupational tasks, theoretical feasibility and actual work-related Claude use, with mode-of-use weighting. It is not an unemployment probability, hiring forecast or measure of an entire job being automated.
We present the dataset's highest ten and 40 additional selected, sourced examples—not an invented global top-50 ranking. We keep U.S. occupation labels and observation dates attached.
Anthropic research ↗ Underlying CSV ↗Scientific claims have stages.
Computer predictions are not experiments. A company mathematical proof announcement and an institute's assessment have distinct review processes. Phase 2a drug studies are not drug approvals. Promising aging biomarkers are not proof of lifespan extension. We state what has not been demonstrated.
The publisher and the observation date travel together.
Each referenced breakthrough is linked to its original source and dated. A data refresh or website visit does not change the original observation date. Market indicators and labor surveys are not labeled 'live' merely because a website refreshes.
Public website is isolated from the studio.
The public-facing site has no private operator controls, tokens, local API connection, database write endpoint, remote administration or telemetry collection. No part of the public website makes a request to the Mac's localhost server.
Corrections are explicit.
When an occupation rank, scientific milestone, event or source is corrected, we revise the underlying record and update the editorial checkpoint. Original-source accuracy is prioritized over viral phrasing.