Potential violations of the Llama 3.2 Community License Agreement

[murawaki]

First of all, great work! I have also worked on linguistic steganography. It is a shame that I have not published a working demo for embedding texts.

I noticed that you are using the Llama 3.2 1B Instruct model, which comes with a tricky license. Specifically, the Acceptable Use Policy prohibits:

3. Intentionally deceiving or misleading others, including:
   e. Representing that the use of Llama 3.2 or its outputs are human-generated.

Since linguistic steganography inherently involves making LLM outputs appear human-generated, this could put us at risk of violating the license.

I raised a similar concern about the BLOOM Responsible AI License.

[shawnz]

Thank you for the notice about this, I added a note to the readme to let potential users know about this restriction.