AI-generated lure document with no spelling errors

Google has sued a smishing network allegedly using Gemini AI to craft phishing messages — the first major case of an AI model weaponized at scale for credential harvesting by organized crime. AI-generated lure documents are hard to spot because they lack the spelling errors and awkward phrasing that traditionally betray phishing, and they arrive as docx, PDF, or eml attachments, a spearphishing pattern mapped to T1566.001. This demo inspects a synthetic phishing document containing no real credentials or links to live infrastructure. OPSWAT AI Content Inspector analyzes the text for machine-generation patterns and clearly flags the document as AI-assisted social engineering, giving defenders a new detection signal.