ARIC Insights | How LLM watermarking actually works

Anthropic recently announced that Claude will embed an invisible watermark in the text it generates – a move that sparked heated debate on social media. The underlying technology, however, is not new: methods for watermarking LLM output have been studied and published for years. In this talk, Pascal explains how text watermarking actually works, and brings some facts to an emotional discussion.

Details

Donnerstag, 10.09.2026 um 10:00 Uhr

Veranstalter

Artificial Intelligence Center Hamburg (ARIC) e.V.

Anmeldung zur Veranstaltung