What are the Claude "indications"?

Because there's a danger now that any writing (human or otherwise) can be labeled LLM-produced. So we need accurate heuristics, or none at all.

Looks like probably not Claude based on their privacy/terms:

>AI-powered feedback

>Storica uses artificial intelligence (OpenAI) to provide feedback on your writing. Your written content is sent to OpenAI's API to generate corrections and suggestions. We do not use your writing content to train AI models. Your writing is processed solely to provide you with immediate feedback.

It does look LLM-generated though.

The copy on the main page gives a first clue. "A daily reading club for language learners" available on 7 languages, on a .club domain...

Now, look at the number of books on each language. Does it sound reasonable that a no-name startup with no contact details (except an email to an aktivlang.com domain that redirects to storica.club) will invest in a serious effort to human-translate and adapt that many books to language learners, without anyone noticing?

I'm not going to argue about the writing style because we know it is an arms race. But look at the underlying business, that business would not exist without AI generated content.

Many (human) translators use DeepL (https://www.deepl.com) in their daily jobs. It employs deep learning (hence the name), but not necessarily LLMs. And it's way better than Google Translate for many languages.

So this offers an alternative, legitimate, route to exactly the website content you're suspicious of.

Staccato sentences, em dashes, word choice, the headings, "X is not Y. It's Z.":

> The legacy of the book has almost nothing to do with the satire. It has to do with the language.

Or this also reads very LLM like to me:

> None of the cruelty is gratuitous, exactly. It is dramatised exhaustion with the genre.