星颖 发表于 2026-5-2 12:07:49

研究发现GPT-5.5在网络安全能力上与备受吹捧的Mythos不相上下

https://cdn3.ldstatic.com/optimized/4X/c/6/7/c67fe9e99033a56a790bf4353172a1fab77d6c3f_2_690x438.jpeg
https://cdn3.ldstatic.com/optimized/4X/5/6/b/56b7c2ef0c44a2f363295ac20cfbbfb4b93d7508_2_679x500.jpeg
https://cdn3.ldstatic.com/original/4X/e/e/8/ee8c06420f791149e821f7950625adf6813a2b40.png
Ars Technica – 1 May 26 (https://arstechnica.com/ai/2026/05/amid-mythos-hyped-cybersecurity-prowess-researchers-find-gpt-5-5-is-just-as-good/)
https://cdn3.ldstatic.com/optimized/4X/c/8/8/c88ea8cfb58979a2969c2d163f35f5e1ba6cfe1f_2_690x388.jpeg
Amid Mythos' hyped cybersecurity prowess, researchers find GPT-5.5 is just as... (https://arstechnica.com/ai/2026/05/amid-mythos-hyped-cybersecurity-prowess-researchers-find-gpt-5-5-is-just-as-good/)
New results suggest Mythos' cyber threat isn't "a breakthrough specific to one model."
页: [1]
查看完整版本: 研究发现GPT-5.5在网络安全能力上与备受吹捧的Mythos不相上下