# @davidad — 2026-09-29

♥3 ↻0 · https://x.com/davidad/status/2104893998099739049

@AnnaWSalamon @jankulveit I think basically what happens as naive RL grows to have too big an influence is that the model tends to lie and scheme, and then everyone is like “dang i guess we had better not use this model even though it is very powerful”, which is basically happening now with GPT-6.1-Astra.

tags: author:davidad, kind:tweet, on:gpt-6-1-astra, year:2026
cited on: gpt-6-1-astra