Overview — AI Safety & the Frontier
AI Public Record: Frontier Voices
Pilot / review copy. This Space is a working demonstration of a source-grounded way to trace how prominent AI figures have publicly described safety, governance, deployment, openness, and the threshold for slowing frontier progress.
What this pilot contains
Six profile pages built from completed Deep Research reports, plus a comparative synthesis. Each profile is a longitudinal record rather than a verdict: it separates the person’s own statements from co-authored or institutional positions and from the report’s interpretation of change over time.
The initial focus is Dario Amodei, whose September 2026 argument that the frontier should be paced is the immediate editorial hook. The companion pages make it possible to test whether that position is unusual, widely shared, or different in its evidentiary and policy logic.
How to read the profiles
Every profile preserves the report’s thesis, chronology, periodisation, thematic analysis, tensions and bibliography. Direct source links are kept in the chronology and bibliography. The reports use evidence-strength labels and attribution cautions; those labels matter because a signed collective statement, a company document, and a personal essay do not establish the same thing.
The reports record public assertions and predictions. They do not establish private beliefs, motives, or whether a forecast came true. The working copies also remove ChatGPT-only citation markers that cannot travel outside the research interface; this does not replace the direct source links that should anchor publication.
Profile pages
| Figure | Coverage | Open profile |
|---|---|---|
| Dario Amodei | 2016–September 2026 | Dario Amodei |
| Sam Altman | 2015–September 2026 | Sam Altman |
| Demis Hassabis | 2010–September 2026 | Demis Hassabis |
| Yoshua Bengio | 2010–September 2026 | Yoshua Bengio |
| Ilya Sutskever | 2010–September 2026 | Ilya Sutskever |
| Geoffrey Hinton | 2010–September 2026 | Geoffrey Hinton |
Comparative view
Read the cross-profile synthesis
This collection is intentionally flat: the pages are peers in the Slate hierarchy. “Dario first” is an editorial ordering choice, not a special page type.
Review questions before publication
- Are the strongest claims supported by direct, accessible primary sources?
- Are personal views, institutional positions, co-authored positions, and interpretation visibly distinct?
- Should the pilot publish as a complete collection, or should the Dario profile be edited into a standalone article first?
- Which paragraphs deserve Slate evidence annotations in addition to the source lists?