A tool for concealing writing style to evade linguistic fingerprinting
from richard547@lemmy.world to privacy@lemmy.ml on 08 May 19:57
https://lemmy.world/post/46595407

Source code and details: git.anarchists.space/nemesis/Lingunymous

Everyone has a unique writing style, word choice, sentence structure, punctuation habits and that can be used to identify them. Lingunymous rewrites your text using a local LLM to mask these patterns, making it harder to attribute writing to a specific individual.

Use cases include:

#privacy

threaded - newest

Kirk@startrek.website on 08 May 21:40 next collapse

Clever! The rare actually useful LLM!

leoj@piefed.zip on 08 May 22:31 collapse

and hallucinations just add value!

racoon@lemmy.ml on 08 May 22:38 collapse

This kind of hallucinations is typical of the LLM at IP 168.234.98.1

phoenixz@lemmy.ca on 08 May 22:59 next collapse

Useful LLM’s, always a welcome sight

its_kim_love@lemmy.blahaj.zone on 08 May 23:46 next collapse

Oh yay now people will sound like soulless AI too!

carotte@lemmy.blahaj.zone on 09 May 00:53 collapse

This is not just the solution to a non-issue for most, and a way to normalize agentic writing — it’s both, rolled into one. 🔥🚀

/j

tixnou@feddit.cl on 10 May 03:48 collapse

This doesn’t just delve—it offers a unique insight. or something idk

SocialistVibes01@lemmy.ml on 09 May 00:07 next collapse

So, a scam to take my writing quirks to make your AI look more human?

TherapyGary@lemmy.dbzer0.com on 09 May 02:05 next collapse

local llm

SocialistVibes01@lemmy.ml on 09 May 02:15 collapse

That explains why you’re calling that a good thing

akwd169@sh.itjust.works on 09 May 02:13 collapse

No one wants YOUR writing because you cant even READ

It says LOCAL LLM right there

chicken@lemmy.dbzer0.com on 09 May 03:05 next collapse

I’ve heard about this sort of technique being used, but an important step I usually hear about isn’t present here, which is to first translate the text to another language before translating it back. Simply asking a LLM to

Rewrite the user’s text so it sounds natural and human-written. "

"Preserve the original meaning exactly.

seems likely to leak things like your word preferences and grammatical quirks, a 7b model isn’t going to be super creative about this and will want to take your lead on things. There needs to be an initial layer of stripping your statement of what makes it unique.

davel@lemmy.ml on 09 May 06:48 next collapse

You can tell pretty much any LLM to talk like a pirate or write like a cop, etc. But using a local LLM (which this one is) is the way to go if you’re serious about anonymity.

j5y7@sh.itjust.works on 10 May 00:11 collapse

Like Yoda we should talk. Easier, it would be.