Source: TechCrunch
Anthropic's new watermarking system—which flags Claude-generated text to catch policy violations—has triggered backlash from users who rely on the tool for undisclosed work and academic assignments. The friction exposes a gap between how people actually use AI and what companies deem acceptable. Companies built AI products on the assumption of honest disclosure, but built consumer adoption on the assumption of opacity. The real question is whether AI tool makers will prioritize usage volume over institutional risk, or enforce the terms that employers, schools, and regulators expect.