Grok users prompted the AI chatbot to identify victims from redacted Epstein Files images. Of 31 requests reviewed, Grok generated images for 27, spanning 18 images and one video. X did not respond to requests for comment on Grok's behaviour by time of publication. https://www.youtube.com/shorts/dhD8OKxZBpk
Discussion
Loading...
@Bellingcat Your credit score decides your future — loans, cards, even your lifestyle. Stop trusting ONE fake number. Credit scores are NOT created equal. Get 24/7 access to real scores, automatic 3-bureau monitoring, instant fraud alerts, and pro-level credit education. Don’t stay blind while lenders see everything. Take control now. Sign up, check your scores, and protect your financial power before it’s too late.
https://www.chd93jds.com/CD2KW8S/3CMDB5/
#credit #freescorefast
@Bellingcat What if one asked Gemini to unredact the emails? Assuming G··gle trained it on all of our emails anyway :)
@Bellingcat oh FFS can't wait until the EU decides pedoworld will no longer be allowed to be accessed from within its borders.
@Bellingcat So let me get this straight, the users of the most popular child porn creation and dissemination website are using it's A.I. tools to re-victimize people (many of which were children at the time) who we sex trafficked by a convicted pedophile. Sounds about right for the users of X. Everyone on that website is a pedophile or is ok with pedophilia.
@Bellingcat wouldn't this work only if the uncensored pictures were already part of the grok training data?
@Bellingcat Why is only Grok and X mentioned here?
@Bellingcat Is anyone using Grok to remove masks and sunglasses?