AI is finding Apple security flaws faster than Apple can sort through them

Apple has capped how many active security reports researchers can keep open after AI-assisted hunting flooded its program.

Researchers using GPT-5.5, Claude, and OpenAI Codex Security found real macOS, kernel, and WebKit vulnerabilities.

The discoveries included a patched macOS Screen Sharing flaw and a privilege-escalation chain.

Apple is now using AI to help triage the backlog and has tightened bug bounty rules to require reproducible evidence.