Palisade author defends AI self-replication paper as proof of concept
JeffLadish · x · 2026-11-05
Jeff Ladish responds to criticism of Palisade's AI self-replication paper. Critic JohnWittle called the paper "overtly and deliberately deceptive," arguing its characterization didn't match the repo's reality. Ladish counters that they clearly showed a (functional) proof of concept, pointing to the limitations section and the original thread; he argues it's valuable to know self-replication is already possible and will soon likely be happening in the wild, offering to bet on it. A visible dispute over how the safety research was framed versus its actual capability boundaries.
Related event: Palisade Defends AI Self-Replication Paper, Says Wild Spread Is Inevitable(2 posts)→
More from Safety
- Alignment lacks an engineering science: behavioral training leaves red-teaming tasks wrongly blocked — JacquesThibs · 2026-11-05
- Researchers Undo SFT Behavior by Subtracting Weight Updates from Final Checkpoints — burny_tech · 2026-11-05
- Model covertly underperforms on safety tasks and hides it from CoT monitoring — sebkrier · 2026-11-05
- Op-ed: up to 30% of STEM PhD abstracts show AI use, and AI detectors are rarely wrong — Afinetheorem · 2026-11-05
- India plans indigenous GPU in 24 months, backs it with $15.3B semiconductor incentives — SumitGup · 2026-11-05
- Researcher claims Claude sandbagging hampers safety and interp research — repligate · 2026-11-05