Observed arrival · 2026-11-06
AI Capabilities Observatory
A public briefing site that organizes claims about current AI abilities into capability assessments and risk scenarios.
- For
- Readers tracking current AI capabilities and risks
- Worth noticing
- Its featured scenario marks 6 of 9 pieces as seen, while listing self-funding swarms and broad service abandonment as not yet seen.
Field notes
The homepage separates capability entries—hacking, agent coordination, coding, research automation, math, and making money—from scenario briefs assembled from those capabilities. Its featured scenario breaks the “broken internet” possibility into nine pieces, reporting six as seen and naming three it says have not occurred. A “How we assess” link, corrections policy, and revision log are listed, but the extracted page also contains lorem ipsum and author/date placeholders; the underlying assessment method is not visible here.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue