IAS英国人工智能安全研究所：2025年国际先进人工智能安全科学报告（英文版）

Between the end of the writing period for this report (5 December 2024) and the publication of this report in January 2025, an important development took place. The AI company OpenAI shared early test results from a new AI model, o3. These results indicate significantly stronger performance than any previous model on a number of the field’s most challenging tests of programming, abstract reasoning, and scientific reasoning. In some of these tests, o3 outperforms many (but not all) human exper