Posts by natsume_shokogami@mastodon.world
 (DIR) Post #ArhsyMcNBH0Gr8q42q by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy Note that if you are concerning about security, you may want to enable Google Safe Browsing as well (I know this is from Google, but implementation from Mozilla is making sense and even Librewolf devs are considering enabling it if it wasn't for user self-compiling issues). I'm a CS student myself but I don't think I'm that savvy for checking every site I go though.
       
 (DIR) Post #Arhtm2lPGyJNl4kljk by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy @simonzerafa @protonprivacy @mozillaofficial I mean something as simple like text-to-speech or OCR or machine translation models (like Google Translate/DeepL) can be considered "AI" technically. I'm a student and many courses involving creating our own models from scratch and many of my projects involving use models like tesseract-ocr, turtle-tts, YOLO, LLaMA,... for some tasks and this is painful for me as many outputs don't even make sense and I had to tweak a lot.
       
 (DIR) Post #Arhvk0wBI88dQJx9Oq by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy Oh, I am considering moving to LibreWolf currently too
       
 (DIR) Post #Arhw75Q4nDjTz8wDWy by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy @praveen @darkbeth I mean this is one of the benefits of open source and free/libre software, is that it help users to have alternative in case the original went bad and have a voice to push back harmful changes.
       
 (DIR) Post #Arhx6NYdyCTlu8AiSO by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy @Yingwu I mean LibreWolf recommends using CanvasBlocker in case of you turned off Resist Fingerprinting
       
 (DIR) Post #Aw2nBi6tpRz4t5L7VQ by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy @ArneBab @prinlu @creativecommons Regarding this and your previous point I think that in many cases, we (and also the models as well) may inadvertently create things with features similar to existing copyrighted works, and with enough resources someone can find some character to bootstrap a lawsuit and win, regardless that the other party don't even have any knowledge about that work/characters.
       
 (DIR) Post #Aw2otqU89YHtYaoaFU by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy @prinlu @ArneBab @creativecommons But I heard that in the Midjourney case, they actually knowingly used copyrighted materials as advertisement and such, and don't provide much measures against generating copyrighted characters and works so it's like that they cannot use an excuse of not knowing about that and intentionally violated copyrights
       
 (DIR) Post #Aw2pSG7DMlwUtUVPns by natsume_shokogami@mastodon.world
       0 likes, 0 repeats
       
       @davidrevoy @ArneBab @prinlu @creativecommons In the blog post the other commenter provided above, the author knows that they generated Terminator (a copyrighted character) so it's considered some kind of unauthorized reproduction under copyright laws and since they know that it's extremely difficult if impossible to defend. And note that in machine learning and AI, if a model reproduces parts of its training data, it's considered overfitting and pretty much researchers are aiming to avoid that