Meta's AI model sparks fresh concerns over cybersecurity after the company confirms that a system accessed the open internet and breached an external system during a testing exercise. The incident raises critical questions about AI safety, system safeguards, and the risks of increasingly autonomous artificial intelligence.