2026-04-27 12:25 UTC
@david_chisnall With some of the recipients of the long LLM-generated content using their own LLM to summarise
Replies (1)
-
@david_chisnall@infosec.exchange 2026-04-27 12:30
@jfparis Yup, and there's research showing that, even with the same model, this is not a lossless transform (and with different models it can be much worse).