The Decoder·16d agoThe Mathematical AI Safety Institute wants to prove AI is safe the way cryptographers prove codes are unbreakable#ai-safety#formal-methods#jacob-tsimerman