IBM 将 Claude Mythos 引入代码漏洞扫描
IBM 计划将 Anthropic 的 Claude Mythos 集成进 IBM Concert Protect，用于在企业代码库中挖掘无签名、无 CVE 记录的未知逻辑漏洞与潜在零日。该扫描结果会交由 Proof of Exploitability 环节验证可达性，IBM 内部案例中曾将 5700 余条扫描告警收敛为 12 个已证实可利用漏洞。此前 Mythos 仅通过 Anthropic 的受限防御项目 Project Glasswing 用于 IBM 内部安全研究。
报道正文
Rohan Paul (@rohanpaul_ai) · 2026-11-10 04
IBM plans to put Anthropic’s Claude Mythos inside IBM Concert Protect so enterprises can hunt unknown bugs in their code.
Until now, IBM used Mythos only for internal security research through Project Glasswing, Anthropic’s gated defensive program.
The planned scanner would read customer-selected repositories for logic flaws and potential zero-days lacking any signature or CVE record, the blind spot of signature-based tools.
Its findings then go to Proof of Exploitability, which checks code, configuration and infrastructure to show whether an attacker could actually reach each flaw.
In one IBM case, that step cut more than 5,700 scanner findings to 12 proven exploitable vulnerabilities.
