tech
Tech analyst Ben Thompson dismisses the 'clearly absurd' concept embedded in AI watermarking
Thompson took particular issue with Anthropic adding a watermark even if Claude was used only for proofreading.
TL;DR
- Tech analyst Ben Thompson criticizes the EU AI Act's transparency requirements, specifically watermarking AI-generated content.
- Thompson argues that watermarking is overburdensome and unlikely to be effective in distinguishing AI content, questioning the definition of 'fake'.
- Anthropic's plan to add watermarks to Claude-generated content, even for proofreading, is seen as problematic due to the difficulty in distinguishing usage.
- OpenAI also plans to expand watermarking to text output, using tools like Google DeepMind's SynthID for images and audio.
- Thompson believes watermarking unfairly credits AI with ideation, comparing it to a pen advertising itself as the author, and insists the focus should be on the human user.