tech

Some Claude users are mad that Anthropic's new watermarks will catch them using it at their jobs, classes

Is Anthropic's new watermarking system a travesty? Some have taken to social media to complain that it is.

Some Claude users are mad that Anthropic's new watermarks will catch them using it at their jobs, classes

TL;DR

  • Anthropic is now watermarking Claude's outputs with invisible code to identify AI-generated text.
  • This policy aims to comply with the EU AI Act's Transparency Code.
  • Some users on Reddit express outrage, viewing the watermarks as a "digital tattoo" that will catch students, journalists, and writers.
  • Critics argue that the watermarks are unethical, especially if they believe they did most of the work and Claude was just a tool.
  • Others defend the watermarking as a necessary measure for transparency and to mitigate risks associated with AI-generated content.
  • A counter-argument suggests hypocrisy, given that AI models are trained on vast amounts of existing data.