AI Model Drives Simultaneous Quantum Cryptography Breakthroughs

Following a recent Simons Institute presentation on an unresolved question in unclonable encryption, MIT PhD candidate Seyoon Ragavan directed OpenAI's GPT-5.6 Sol Ultra through iterative sessions that produced a full proof and draft manuscript. Simultaneously, UC-Santa Barbara professor Prabhanjan Ananth and UCLA professor Amit Sahai employed a custom UCLA interface to the same model, which generated the construction and central proof ideas that they later refined.

Both groups posted their findings to arXiv on the same day, explicitly crediting the AI for the breakthrough. Their papers assert an efficient, assumption-free scheme for unclonable encryption, extending the 2019 framework introduced by Anne Broadbent and Sébastien Lord. The rapid, overlapping results illustrate AI's expanding role in theoretical research and spark debate over attribution and the future of graduate-level work. The authors are now considering merging their results for a conference submission.