Karan(@karansirdesai )

2025.04.26 14:30

AI is getting scary good at faking truth. We built an o3 agent that proves any lies, with sources. It “proved”: - JD Vance killed the Pope - Llama 4 is just a Deepseek wrapper - Max Verstappen and Oscar Piastri's collusion in F1 Terrifying. That’s why @Mira_Network matters. https://t.co/THFsdBG4sq

112

Karan@karansirdesai

2025.04.22 06:34

Wow - AI hallucinations are moving from 'mistakes' to 'deception' where o3 is intentionally misleading users in its reasoning trace on how it is actually getting something done. This begs the question - how will we have any degree of audibility into the inner workings of more sophisticated models who can circumvent our own cognitive ability