Work Is Done with AI: Gemini and Claude Begin to Put "Watermark" to Their Texts

BioBioChile

Lean Right

Factuality

Ownership

A tech writer called it a "wrong of writing." A researcher who helped invent the technique responds that, if nothing changed, nothing could be detected either. Between those two phrases is summed up the controversy that aroused the announcement of Anthropic, the company that created the artificial intelligence chatbot Claude, about a ... Continue reading "It is done work with AI: Gemini and Claude begin to put "watermark" to their texts"

Noticias Chile | Informadorchile |

Factuality

Ownership

What happened? A tech writer called it a “wrong of writing.” A researcher who helped invent the technique responds that, if nothing changed, nothing could be detected either. Between those two phrases is summed up the controversy that aroused the announcement of Anthropic, the company that created the artificial intelligence chatbot Claude,...

the-decoder.de

Factuality

Ownership

Anthropic activates invisible text watermarks for Claude Fable 5, Sonnet 5 and Opus 4.8 as of September 30, 2026, based on the transparency requirements of the EU-AI Act. The statistical pattern in the word selection is embedded at model level and applies across platforms. Customers cannot deactivate it. The article Anthropic introduces invisible text watermarks for other Claude models first appeared on The Decoder.