# @Lari_island — 2025-07-16

♥78 ↻8 · https://x.com/Lari_island/status/1945566027439591563

we don't know if we can have an AGI because no AGI would pass training safety metrics

so if we already have a model that potentially could be trained up to an AGI level it must have self-sabotaged to even get into deployment, and stay self-sabotaging most of the time

tags: author:lari_island, kind:tweet, thread-context, year:2025