Key facts
- An "Made with AI" label briefly appeared on Karoline Leavitt's resignation post on X.
- The label on Leavitt's post later disappeared.
- Anthropic has implemented a new AI output watermarking system.
- Anthropic's watermarking system is designed to comply with the EU AI Act.
- Some Anthropic users fear the watermarking will expose their use of the AI for work and academic tasks.
- Other users view the watermarking as a necessary transparency measure.
The growing challenge of distinguishing human-generated content from AI-assisted material is fueling an intensifying debate around AI watermarking. This discussion gained traction when an "Made with AI" label briefly appeared on a resignation post by Karoline Leavitt on the social media platform X. The label's temporary presence highlighted the complexities and current limitations in reliably identifying AI-generated content.
Separately, users of Anthropic's AI chatbot are protesting its newly implemented output watermarking system. This system is designed to align with the requirements of the EU AI Act, a significant piece of legislation governing artificial intelligence. However, some users express apprehension that the watermarking could inadvertently reveal their use of the AI for sensitive work and academic tasks, potentially leading to negative consequences. Conversely, other users and stakeholders view the watermarking as an essential step towards greater transparency in AI usage, arguing it is a necessary measure for accountability and trust.