Hmm it's nothing special it's as Jensen Huang puts it " a modality grounded model", openai needs the model accessing as much information and the internet is done, they need to push their model dynamically and that needs it exploring the world(because it doesn't need reasoning or learning like self driving cars that are no where near level 5 or 3 in level of intelligence, but definitely above cruise control)but openai needs dynamic information and allowing people access it via their phones helps accelerate cultures, behaviour, people, it's the next stage to supercharge the next evolution on ai connecting outside the chat box.
It will normalise biometric facial recognition in 5 or so years, right now people would scream at it, but I suspect openai would block faces and give it push by request not listen by demand.
But the retrieval based information from 1 billion people will supercharge it and it'll be free for open ai due to you freely agreeing to to terms naively.
Think about it, openai entire business model is collecting information, without it's nothing, if openai hired 50,000 people to go out and explore the world for the llm model to gradually grow Google would have left openai for dead.
Openai has made a strategic strategy getting this next stage of ai out, but for free while removing openai from legalitys on copyright clauses, and more people onboard within less then 5 months will allow openai bring out chatgpt5 under Nvidias Blackwell h200.