Dead startups fuel AI training with their data

artificial intelligence labs are acquiring data from failed startups to train their models, as they exhaust public and synthetic datasets. companies like keelan miskell's former crypto startup are selling their digital archives for tens of thousands of dollars. this growing market raises legal and ethical questions about data ownership and usage. it affects the development of ai agents and the startup ecosystem, as buyers seek real-world records to improve ai training. this trend may change how startups approach data management and ai labs approach model development.