←back to thread

108 points liquid99 | 1 comments | | HN request time: 0s | source
Show context
gryfft ◴[] No.43359090[source]
It's been mentioned elsewhere recently but this presents an accessibility nightmare for screenreaders and similar assistive technologies.
replies(4): >>43359382 #>>43359523 #>>43359799 #>>43360499 #
itake ◴[] No.43359382[source]
but great for fraudsters trying to side step content moderation models!
replies(3): >>43359439 #>>43359657 #>>43359658 #
waltbosz ◴[] No.43359439[source]
Are these models really able to be fooled by text tricks like this?
replies(1): >>43359713 #
1. itake ◴[] No.43359713{3}[source]
It depends on what you mean by "models".

LLMs? No. But LLMs are too slow for content moderation at scale.

Custom trained models? Maybe. Is the unicode characters in the training data?