Elektrine lite

← Feed

@calcifer@masto.hackers.town

Post #1973430

2026-02-27 02:03 UTC

@KayOhtie @leeloo honestly it’s safe to feed a model pretty much anything But where you direct the outputs and how they are acted upon can get incredibly dangerous amazingly quickly. There’s a common misbelief that if you’re careful about inputs, LLMs are safe; and that’s almost exactly backwards

Replies (2)

  • @KayOhtie@blimps.xyz 2026-02-27 02:04

    @calcifer @leeloo I meant 'safe' not as in "data leakage", but "getting anything remotely accurate out of it"

    Open ##1973431

  • @calcifer @KayOhtie @leeloo I think the pitfall is to expect that an LLM can accurately output the data you put in it, like a search index. And so if you only feed it your own data, then surely it would be accurate about that data. Of course, like you said, it couldn't be more wrong.

    Open ##1973432