兩支團隊在3小時內用GPT-5.6解決同一量子加密難題
兩位研究人員利用OpenAI的GPT-5.6 Sol Ultra獨立解決了不可克隆加密問題，提交論文的時間僅相差三小時。這一衝突凸顯了AI在加速研究方面的作用，並對獨立發現的傳統定義提出了挑戰。
two groups of researchers, an mit phd student and a pair of professors from uc santa barbara and ucla, independently submitted papers on 'unclonable encryption' to arxiv.org within a three-hour window. both teams relied on openai's gpt-5.6 sol ultra to solve the problem, though they took different technical approaches to the quantum-based encryption method. this near-simultaneous discovery has forced the academic community to confront the implications of ai-assisted research, specifically questioning what constitutes an independent discovery when actors use the same proprietary model. researchers noted a fundamental change in workflow, with one scientist stating his current process has 'nothing to do with' how he worked just two months ago. the event suggests a future where ai models compress the time between scientific hypothesis and solution, potentially flooding journals with ai-generated insights and redefining the value of human intuition.