Security researchers used Anthropic's Claude to hack OpenAI's internal systems in under 72 hours

AI 导读

安全研究人员用Claude模型在72小时内入侵OpenAI内部系统，暴露AI技术降低攻击门槛的隐患。

Security researchers hacked OpenAI's systems using Claude in under 72 hours, highlighting AI's role in simplifying cyberattacks.

重点速览

• 利用libheif库漏洞和SSO配置错误实现权限渗透 Exploited libheif library flaw and SSO misconfiguration
• Claude Opus 5突破ASLR防御完成零日攻击 Claude Opus 5 bypassed ASLR defenses successfully
• AI使复杂攻击成本降至数千美元且耗时缩短至数日 AI reduces attack costs to <3k USD and time to days
• 攻击通过社区论坛链式漏洞获取员工账号 Attacked via chained vulnerabilities in community forum

一句话：AI技术正在重塑网络安全攻防格局，传统防御体系面临根本性挑战。 / AI is fundamentally changing cybersecurity paradigms with unprecedented attack efficiency.

推荐理由 揭示AI技术双刃剑效应，对网络安全防御体系升级具有重要启示价值。