On August 2, new transparency and disclosure requirements went into effect for chatbots and AI-generated content under the European Union (EU) Artificial Intelligence (AI) Act. Panic or unrest among design engineers surrounding these new watermarking algorithms is somewhat overblown, according to Benoit Tanguay, founder of the medtech startup Auromone. “The shorter and more fact-dense the text Claude produces, the less room the watermark has to show up at all,” said Tangua.
Source: Engineers Tease Out Impacts of Anthropic’s Invisible Watermark