Anthropic Frontier Red Team 评估 GLM-5.3 的二进制利用能力扩散
Quoting Anthropic Frontier Red Team
Anthropic Frontier Red Team 在内部 Binary Exploitation 基准的 100 个任务上评估发现，GLM-5.3 在 4% 的试验中实现完整控制流劫持，Claude Mythos Preview 为 6%，而 Claude Opus 4.6 和 GLM-5.2 均无一成功。作者引用该结论指出高级网络能力扩散的门槛已被跨过。
We evaluate several models on 100 tasks from the [internal Binary Exploitation benchmark] (selected at random), and find that GLM-5.3 develops full control flow hijacks in 4% of the trials; Claude Mythos Preview did so in 6%. Although GLM-5.3 performs below Claude Mythos Preview here, a meaningful threshold has clearly been crossed: earlier models, like Claude Opus 4.6 and GLM-5.2, do not succeed in any of them.
— Anthropic Frontier Red Team, GLM-5.3 and the spread of advanced cyber capabilities
Tags: anthropic, generative-ai, ai-security-research, glm, ai, ai-in-china, llms
来源：Simon Willison 博客 · simonwillison.net