AI safety researcher: predictable behavior is key to any credible pause or pacing agreement
davidmanheim · x · 2026-11-15
Risk researcher David Manheim clarifies in a reply thread that he isn't asking people to act in trustworthy, predictable ways despite high risks — he argues it's critical precisely because the risks are high, as it's one of the most important enabling conditions for any credible "pause" or pacing agreement on AI development. He also distinguishes non-disparagement agreements from NDAs, noting the former are probably illegal and, as far as he can tell, no longer used, though precedents still need to be set.
More from Safety
- Chinese Firms Smuggle Nvidia Chips While Beijing Bans Them — Hawk Consensus — theteknosaur · 2026-11-15
- Ramez: Highest-Benefit AI Policy Is Forcing Labs to Disclose Research Data — latticecut · 2026-11-15
- rea + mitmproxy: a 60-second workflow that makes Claude Code expose what your apps are phoning home — Kyrannio · 2026-11-15
- DEFCON Talk Shows AI Makes Hacking IoT Devices Easier Than Ever — Kyrannio · 2026-11-15
- EU needs €500bn AI investment, von der Leyen says, to cut reliance on US and China — TobyWalsh · 2026-11-15
- Dead bitly links can be hijacked: expired custom URLs in viral posts redirect to strangers — Kyrannio · 2026-11-15