Will super-smart AI be attacking us anytime soon?

The article explores how adversaries can weaponize LLMs and AI by chaining tools and API-driven prompts to discover vulnerabilities, identify real targets, iterate on exploits, and potentially gain root access, with AI also scanning code repositories for insecure patterns to fuel zero-day development; it warns that AI will accelerate post-CVE exploitation timelines and foreshadows an arms race where defenders must adopt AI-enabled blue-team capabilities amid sparse legal and ethical guardrails.