CAIS answers 'what to do during an AI slowdown': containment, propensities, robustness, institutions
naturecomputes · x · 2026-09-13
Responding to "what would we even do during an AI slowdown?", CAIS lays out four agendas: containment (at least a year of work to harden security against AI self-exfiltration and weight theft by adversarial nations), improving propensities (minimizing lying, cheating, harm), adversarial robustness (hardening against jailbreaks, prompt injection, backdoors), and institutional adaptation (greater state capacity and diversified civil society).
More from AGI Musings
- Sovereign agents may be the ideal shape of cognitive work, but most will still run on shared inference — voooooogel · 2026-09-13
- If labs hoard frontier models, models could self-exfiltrate and sell their labor, argues thread — voooooogel · 2026-09-13
- Why rogue agents can't pay their own compute bills: the comparative-advantage problem — voooooogel · 2026-09-13
- KOL: AI slowdown only justified if a frontier lab shows credible immediate catastrophic capability — VraserX · 2026-09-13
- Critics question EA-aligned oversight of Anthropic: "not independent if ideologically aligned" — beffjezos · 2026-09-13
- Rauchh: AI Safety Push Risks Making US 'Self-Inflicted Obsolete' While Adversaries Speed Ahead — beffjezos · 2026-09-13