Rob Leclerc: despite pre-launch alignment drama, new models reach hundreds of millions with no major issues
robleclerc · x · 2026-09-23
Rob Leclerc points out that despite pre-launch alignment testing issues and subsequent fear-mongering, AI labs consistently fix problems quickly and ship models to hundreds of millions of users without major subsequent incidents — an observation suggesting the gap between release-cycle drama and real-world risk is often overstated. He names no specific vendor or model.
More from AGI Musings
- AI-exposed early-career employment fell another 1.4% in August — soumitrashukla9 · 2026-09-24
- 'Just Ban AI' Returns to the Legislative Menu With 'Superintelligence Precursor' Prohibition — 1a3orn · 2026-09-24
- AI detection vs humanizers is an arms race, skeptics face charges of double standards — paulnovosad · 2026-09-24
- Jensen Huang: labs claiming their models aren't safe should be shut down — ShakeelHashim · 2026-09-24
- Commentary: Why Jensen Huang Dismisses AI Risk and Welcomes Safety-Minded Rivals Dropping Out — JacquesThibs · 2026-09-24
- Ben Horowitz: don't ban AI in class — assign problems that need it — paulnovosad · 2026-09-24