News for hackers. Stuff that matters.
Re:Identifying and Eliminating CSAM in Generative ML Trainin... by akomtu on Wednesday December 20, 2023 @05:07PM (#38714281)
Are they going to "lobotomize" ML models to make csam unthinkable for it, or are they going to teach it that csam is bad?
Re:Identifying and Eliminating CSAM in Generative ML Trainin... by tedivm on Wednesday December 20, 2023 @05:22PM (#38714431)
The two paragraph abstract states that the goal is to identify CSAM in the training data itself and remove it. They ran a study and found "many hundreds of instances of known CSAM in the [LAION-5B] training set".
Re:Identifying and Eliminating CSAM in Generative ML Trainin... by GaggiX on Wednesday December 20, 2023 @05:13PM (#38714330)
They are going to report the content to the CDNs and try to remove the entry URL from the LAION dataset and that's mostly it.