All Major AI Models Can Generate Phishing Attacks—And It May Be Hard to Stop

Tests of five leading AI models—Claude 3 Haiku, GPT-4o, Nemotron, DeepSeek-V3, and Qwen—found all could produce convincing phishing content. Analysis: The pattern across different models suggests this capability is built into how large language models work, not a side effect of training choices. The challenge: blocking it without crippling legitimate uses.