(github.com)

536 points BlueFalconHD | 2 comments | 06 Jul 25 19:50 UTC | HN request time: 0.636s | source

I managed to reverse engineer the encryption (refered to as “Obfuscation” in the framework) responsible for managing the safety filters of Apple Intelligence models. I have extracted them into a repository. I encourage you to take a look around.

1. apricot ◴[06 Jul 25 23:11 UTC] No.44484987[source]▶

>>44483485 (OP) #

Quis custodiet ipsos custodes corporatum?

replies(1): >>44490311 #

2. tempodox ◴[07 Jul 25 13:43 UTC] No.44490311[source]▶

>>44484987 (TP) #

nemo videtur.

↑

I extracted the safety filters from Apple Intelligence models