Hacker News Reader

They don't do it for text I think that is the core issue.

Somehow it is acceptable for us to have photo, video or audio watermarked but not text or worse code.

Realized your right about OpenAI but they plan to. In the link they state "our goal is to expand provenance signals to all modalities including text" so its comming up regardless. I pretty sure Google does it though.

That’s what I can’t put my finger on, see my other comment : why would Anthropic tell the public.

The EU AI Act requires them to do this. It would be pretty weird for them to say "we're not going to do this" while secretly doing it anyway.

Thanks

I don’t think they are:

> One need look no further than Reddit to find evidence of brewing discontent, though other posters on the site are not in agreement.

It’s a post about some Reddit comments, a nothing-burger.