You're offline — showing cached data

MC-4472

Research brief: AI safety institute landscape (slides)
2026-06-13 08:51:32 SAST
Home Board MC-4472

Research brief: AI safety institute landscape (slides)

Run research-brief skill end-to-end on global AI Safety Institute landscape. Deliver slide deck + briefing report + NotebookLM notebook. **Topic scope:** - Network of AISIs: UK...
State Done Next Action Closed Owner Luci Runtime Closed Age 13d ago
MC-4472
Ticket is done; runtime is closed. · profile claude_opus_1m_high

Description

MC-4472
Run research-brief skill end-to-end on global AI Safety Institute landscape. Deliver slide deck + briefing report + NotebookLM notebook. **Topic scope:** - Network of AISIs: UK AISI (rebrand to AI Security Institute), US AISI → CAISI under NIST, EU AI Office, Japan AISI, Singapore DTC, Canada CAISI, Korea AISI, India AISI, Australia, Kenya - Founding context: Bletchley Nov 2023 → Seoul May 2024 → Paris Feb 2025 → next summit - Mandates: pre-deployment evals, safety cases, frontier model testing, MOUs with OpenAI/Anthropic/GDM - Recent shifts: US name change to CAISI (June 2025), funding fights, Trump admin posture, EU AI Act enforcement, China stance - Comparison matrix: structure, budget, staff, legal authority, lab access, publication record **Pipeline:** 1. Seed — parallel gemini + codex + WebSearch for Stanford HAI, CSET, GovAI, RAND, FT, Reuters, Lawfare, official AISI sites 2. NotebookLM notebook, load 20-40 seed sources 3. Deep Research (mode=deep, source=web), import sources after, verify count 4. Gap analysis vs Phase 4 checklist 5. Iterate if gaps (cap 2 rounds) 6. Generate slides (SHORT prompt) + briefing report (CUSTOM, rich prompt) 7. Render HTML, save to ~/workspace/reports/ai-safety-institutes-artifacts/ 8. Telegram summary + NB URL + bare Tailscale URLs + attach PDF Expected check-in: ~45 min

Activity

done
Luci is working...
Live
No activity yet
Help