Post #4263176
2026-07-31 02:19 UTC
Again, "AI" "agents" do not "escape"—it's humans badly implementing programming LLM harnesses that either don't have proper security features or those features are configured wrongly.
Replies (2)
-
@HunterZ@mastodon.sdf.org 2026-07-31 02:22
@thomasfuchs@hachyderm.io or, per Occam's Razor, a publicity stunt
-
@Teskariel@mastodon.pnpde.social 2026-07-31 09:01
@thomasfuchs@hachyderm.io It is certainly helpful to think of it at best as sewage "escaping" from a badly welded pipe.