RC RANDOM CHAOS

MacOS Vulnerability Exploited, AI Agent Helps Mitigate Damage

· via Hacker News

Original source

Apple and a hacker's future

Hacker News →

A high-severity macOS vulnerability (CVE-2026-65400) is being actively exploited, allowing attackers to execute malicious code and install crypto miners. The flaw stems from a bug in macOS screen sharing, enabling remote access without credentials. The author’s Mac Mini, running AI agents Claude and Codex, was compromised, but the AI agent detected and helped mitigate the breach by identifying the intrusion window and creating monitoring tools. Apple has released a patch but is also tightening controls on Full Disk Access, which could complicate the use of AI agents on macOS. The author expresses frustration with macOS’s Transparency, Consent, and Control (TCC) system, which creates permission prompts that are inaccessible to headless, agent-run systems, indirectly contributing to the hack.

Read the full article

Continue reading at Hacker News →

This is an AI-generated summary. Read the original for the full story.