Attribute inference attacks

1. Successive Halving: AI (Brace For These Hidden GPT Dangers)