@stephen@ephemera.vital.org.nz
Post #2345813
2026-04-01 18:18 UTC
Argh. This seems like a very inappropriate use of LLMs to me. Although it's a tender at this stage, not clear that any implementation exists at this point.
https://www.thepress.co.nz/politics/360978799/ombudsman-warns-humans-need-make-final-call-redactions-made-ai-bot
Replies (1)
-
@androcat@toot.cat 2026-04-21 07:55
@stephen@ephemera.vital.org.nz Inappropriate, and just completely unhinged. It will never work. LLM "guardrails" don't actually work. It's a roll of the dice at best, and very hackable.