Apple has imposed limits on AI-generated security reports in its bug bounty programme, the Financial Times reported, citing a flood of AI-enabled vulnerabilities that outpace human review. The company introduced a cap on submissions and a 30-day cool-off period for researchers using its Feedback Assistant tool, as not all reports are hallucinations or theoretical threats—some are legitimate and fixable, but distinguishing them is difficult. Italian cybersecurity firm Bynario said its Atlas platform, powered by GPT-5.5, discovered a macOS Screen Sharing vulnerability allowing an authenticated VNC viewer to access protected data and create files with root privileges; Apple assigned it CVE-2026-43760 and fixed it in macOS Tahoe 26.6. Bynario also reported 50+ vulnerabilities in 3 weeks, including a privilege escalation chain enabling complete control of a Mac. The firm said it found a separate serious vulnerability but could not submit it because Apple had blocked further reports, estimating the flaw's black-market value at up to $200,000. Apple’s own defence relies on similar technology, with advisories crediting researchers using Claude for a kernel vulnerability and OpenAI’s Codex Security tool for surfacing WebKit issues. To sharpen signals, Apple raised its top bug bounty past $5 million for severe exploit chains and introduced "target flags" to prove flaws reach protected system parts.