苹果公司因面临大量由人工智慧生成的虚假安全漏洞报告（所谓的「AI 垃圾」），已限制安全研究人员向其内部安全团队提交错误报告的数量。这些低品质的提交是由业余漏洞猎人使用生成式 AI 工具所产生的，给苹果的审查系统带来了巨大的压力。

义大利资安新创公司 Bynario 利用 ChatGPT 在短短三周内发现了 50 多个 MacBook 作业系统的漏洞，其中包括一个严重的权限提升漏洞。然而，由于苹果最近实施的提交上限和冷却期政策，该公司无法将此重大漏洞回报给苹果。

尽管 AI 产生了大量虚假报告，但它同时也帮助了专业研究人员发现危险的漏洞，甚至苹果本身也利用 AI 来加强其软体安全。这使得漏洞赏金计划的挑战从寻找漏洞转变为如何以机器的速度验证、优先处理及回应这些问题。

Apple has restricted the number of bug reports security researchers can submit to its internal team due to a flood of "AI slop" — hallucinated or low-quality security risks. These submissions, generated by amateur bug hunters using generative AI tools, have placed immense pressure on Apple's review system.

The Italian cybersecurity startup Bynario utilized ChatGPT to discover over 50 bugs in the MacBook operating system within three weeks, including a severe privilege escalation vulnerability. However, the startup was unable to report this critical flaw to Apple because they hit the newly implemented submission caps and cool-off periods.

Although AI has generated a massive volume of fake reports, it also assists skilled researchers in finding dangerous exploits, and even Apple uses AI to strengthen its own software security. This dynamic is shifting the challenge of bug bounty programs from merely finding vulnerabilities to validating, prioritizing, and responding to them at machine speed.