AI has changed both sides of security. This page covers offensive use — AI-assisted phishing, automated vulnerability discovery, deepfake fraud — and defensive use, including AI-driven detection, automated patch generation and security copilots. It also covers the attack surface AI systems create themselves: prompt injection, model and supply-chain poisoning, agent credential leakage and jailbreak research. If you run models or agents in production, this is the feed that tells you what can go wrong.
Quick Answers