The risky part is not a dramatic platform breach; it is the possibility that legitimate AI-assisted web tools can be repurposed to generate convincing brand-lookalike phishing pages at speed.