AI commentator laments frontier labs leak nothing, not even model configs
teortaxesTex · x · 2026-11-02
teortaxesTex wishes for a genuine leak from Anthropic or OpenAI — not necessarily weights, even just a model config would be a breakthrough for community discourse. But frontier labs are, alas, very good at keeping this information under wraps.
More from AGI Musings
- After AI solved the Erdős conjecture, 'just synthesis' no longer holds, argues Berezin — ipeirotis · 2026-11-02
- Reddit forces logins to stop scraping; critics say the content's already been sold — nptacek · 2026-11-02
- "Most people never mastered Excel — AI adoption will follow the same path" — Inner-Schedule8998 · 2026-11-02
- China produced 40,000+ humanoid robots by June 2026; 100,000 full-year figure is only a forecast — ingliguori · 2026-11-02
- Michigan prisons to deploy AI that predicts inmates' future violent risk for security classification — Polymarket · 2026-11-02
- Nadella reportedly has second thoughts on AI alignment: smarter models aren't automatically safer — VraserX · 2026-11-02