Reddit users complain Claude's subtle 'watermarking' word choices make it unusable for copyediting
Gandleon · reddit · 2026-10-22
A Reddit user vented that Claude's new version seems to carry an imperceptible 'watermark' in its outputs — recurring, subtly odd word choices that make copyediting untrustworthy. His example: he mistyped 'unvoluntary' for 'involuntary' (a term of art cited in surrounding text); Claude caught the error but 'corrected' it to 'unchosen'. He says this happens constantly on the new version while the previous Opus wouldn't do it, contradicting claims that watermarking doesn't affect quality.
More from Models
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- DeepLearning.AI weekly: GPT-6.1 hits 97.9% byte parity on COBOL migration — DeepLearningAI · 2026-10-23
- Marker's lite mode scores 75.1 vs 42.3 for Azure Read on OmniParseBench — VikParuchuri · 2026-10-23
- Datalab launches lite OCR at $0.50 per 1,000 pages, 20-30x cheaper than Azure, Google and AWS layout APIs — VikParuchuri · 2026-10-23
- LiveNerf's 29-day data finds no nerfing of Claude Opus 5.5 — TheOnlyVibemaster · 2026-10-23
- PHP Foundation reveals Anthropic's Claude cyber model scanned PHP, validated with OpenAI's Daybreak Red — saibharadwaj · 2026-10-23