Two Research Teams Use GPT-5.6 Sol to Crack the Same Quantum Cryptography Problem and File Papers Three Hours Apart

An MIT doctoral student and two University of California researchers independently used GPT-5.6 Sol Ultra to resolve a six-year-old open problem in unclonable encryption, both prompted by the same conference talk. Their preprints landed on arXiv just three hours apart, raising unresolved questions about how academic attribution should work when AI sharply compresses discovery timelines.