AI safety debate: trustworthy behavior framed as prerequisite for any AI pause agreement
timcolbourn · x · 2026-11-15
In an ongoing exchange with @timcolbourn, safety researcher @davidmanheim argues that acting in trustworthy, predictable ways is critical because risks are high—it is one of the most important enablers of any credible AI pause or pacing agreement. The thread began with criticism of frontier labs' permanent non-disparagement NDAs.
More from AGI Musings
- Microsoft's 15-hour outage: why a senior engineer says AI won't replace programmers — amaarora · 2026-11-15
- Dev-turned-Ops argues Ops roles may be more resilient to AI agents than Dev jobs — cl3don · 2026-11-15
- Ramez: Highest-Benefit AI Policy Is Forcing Labs to Disclose Research Data — latticecut · 2026-11-15
- AI safety researcher models RSI timelines against his kid's school start — HaydnBelfield · 2026-11-15
- Social media will die within 5 years from AI slop and bots, predicts founder Kyle Gawley — kylegawley · 2026-11-15
- After 10 years coding, he was replaced by Claude — so he opened an AI-run marketing agency — Fluffy_Lie6789 · 2026-11-15