**[AI Risk Research Points to Bioweapons and Catastrophic Pathogens]** According to Axios, a joint study by MIT FutureTech and the University of Queensland reveals that among 272 researchers evaluating 24 AI risks, there is a 12% probability that, even with existing mitigation measures, AI's dangerous capabilities could lead to catastrophic outcomes—such as "over 1 million deaths or losses exceeding $100 billion"—by 2030. The probability of AI facilitating weapons and large-scale harm is also 12%, and without mitigation measures, the risk exceeds 20%. The study highlights that AI-assisted development of chemical or biological weapons ranks among the highest risk categories. Recently, several leaders of major AI companies have issued a joint open letter warning about the risks of AI-derived bioweapons, calling for enhanced scrutiny of model capabilities and safety measures, with particular attention to the potential malicious use of open-source models.