AI Model Fundamentally Cracks CAPTCHAs, Scientists Say

The report says the model has defeated tests used to tell humans from bots. It uses reasoning to explain the jumbled letters.

Read more on NPR