The Invisible Ink
Anthropic and other major AI providers now embed invisible watermarks in AI-generated text, per an EU transparency rule that took effect August 2, 2026, and this marking is global, not EU-only. The catch: a detected mark only shows content was processed by AI, not that AI wrote it, so a non-native speaker's human-written draft that got a grammar check can trigger the same flag as fully generated copy.
By Terence Ronson, Founder & Managing Director, Pertlink Limited