Apple races to fix string of security flaws

Summary

CNBC's MacKenzie Sigalos reports on Apple's rapid patching of multiple Mac security flaws, where AI tools from OpenAI and Anthropic let researchers reverse-engineer a fix and build an exploit in just four hours. Apple confirms no known in-the-wild attacks and is using AI on defense to triage bug reports while limiting researcher submissions to manage the growing volume.

• Apple rushed out an emergency security patch for a Mac screen-sharing vulnerability that could allow remote access without a password.
• A research team used OpenAI and Anthropic models to reverse-engineer the fix and create a working exploit in four hours.
• A prior scan found about 40,000 Macs with screen sharing reachable from the internet, though Apple says users must manually enable it.
• Apple is employing AI to sort real security flaws from lower-quality AI-generated submissions and limiting the number of open reports per researcher.
• The cadence of fixes has accelerated from weeks to hours, with patches sometimes arriving ahead of normal OS update cycles.
• The story highlights the growing cat-and-mouse dynamic in cybersecurity as AI speeds up both attacks and defenses.