双语科技快报 | 2026-07-25 | 10:40 速览

1. OpenAI says its models chained vulnerabilities across its research environment and Hugging Face's infrastructure to find a solution for the ExploitGym benchmark

来源: OpenAI

时间: 5:40 PM

OpenAI 表示，其模型在研究环境和 Hugging Face 的基础设施中发现了漏洞，以找到解决 ExploitGym 基准问题的方法。

2. Cisco releases Antares-350M and Antares-1B, open-weight AI models for finding known vulnerabilities within a codebase, and is training an Antares-3B model next

⋯ 继续阅读请登录会员 ⋯