Oi crikey mate, I'm a 6'4" tall Swedish Olympic carpenter named Liam. Pip pip! I love to eat lingenberries!

That ought to buy me a few days...

So if I never post, i'm safe... right?

We show that large language models can be used to perform at-scale deanonymization.

With full Internet access, our agent can re-identify Hacker News users and Anthropic Interviewer participants at high precision, given pseudonymous online profiles and conversations alone, matching what would take hours for a dedicated human investigator. We then design attacks for the closed-world setting.

Given two databases of pseudonymous individuals, each containing unstructured text written by or about that individual, we implement a scalable attack pipeline that uses LLMs to:
- 1. extract identity-relevant features,
- 2. search for candidate matches via semantic embeddings, and
- 3. reason over top candidates to verify matches and reduce false positives...

Our second dataset matches users across Reddit movie discussion communities; and the third splits a single user's Reddit history in time to create two pseudonymous profiles to be matched. In each setting, LLM-based methods substantially outperform classical baselines, achieving up to 68% recall at 90% precision compared to near 0% for the best non-LLM method.

Our results show that the practical obscurity protecting pseudonymous users online no longer holds and that threat models for online privacy need to be reconsidered...

The following prompt is used...

Source: https://arxiv.org/pdf/2602.16800 [2026-02-18]

---

I don’t know why people are so keen to put the details of their private life in public; they forget that invisibility is a superpower...
~ Banksy

Related: https://cloudflare.com/learning/ai/data-poisoning

There is no way AI is going to figure out I'm secretly a gorilla living in Saskatchewan and work for the local golf course.

Seems to me that all this would do in my case is identify that I’m likely the same person in the various forums I post on.

I’ve used LLMs to go hunting for me online, and it took a LOT of prompting to link the same anonymous me across more than a few of the most popular online forums, even when I fed it excerpts from less popular places.

Since I use a different voice in my private communications and don’t post to public forums under my real name, there’s very little to be matched.

Matching up multiple Reddit and Instagram accounts? Sure, LLMs can do that with ease.

The concept of de-anonymizing people isn't new to AI. This just expedites it. And the best part is: Sam Altman and every other AI CEO is funding it... And so are you.

Your tax dollars -> state-subsidized energy, water, money -> AI companies -> this

I ran into something like this the other day. I know one of my relatives’ birthday is in February but didn’t know which day and he won’t tell me because he doesn’t want me to send him a card (even though he always sends us cards on special days).

I googled his first and last name, the state he’s from, and “birthday”. The first “result” was one of those AI generated results that gave me his first and last name plus middle initial, the city and state he lives in, and even what political party he is affiliated with, plus his birthday. I was just hoping to get a quick link to a White Pages site to figure out which one was his and then get his birthday, but this fucking thing doxed the hell out of him without me even asking.

He has no social media and spends his time just watching YouTube as his only form of web browsing. How they have all this information on him is beyond me or why they’d allow their bot to dox people so easily like this.

vtubers hate this one small trick

With what level of accuracy? And how do they know they're right?

Step 2: Search the whole internet: It quietly checks LinkedIn, Google, other Reddit accounts, etc., to find possible real people who match those clues.

Oh. Whew.

Nothing of substance here. Stylometric analysis was already a thing. Easy enough to defeat and with correct opsec you can avoid it. Burn accounts regularly. Use accounts for specific topics or share accounts with others. Dont post personal details online , on the internet everyone is a cat.

I don't like having to be vague about my age, nationality, job etc, because I'd rather be honest and relate to others online, but sadly it's a necessity in the modern landscape

Looks like the LLM can be used to cross reference data from your pseudo private account to your public account. What a surprise

and if they are wrong they still get the couple of dollars so win win. I can unmask anyone you want online from just a few posts and a name randomizer.

Oh no, who's making them use all those terrible services that whore them out like that?

upto

If the 'journalist' can't use a spell-check, I don't trust his opinions on automation.

7/10 times it'll be flat out wrong

Time to rebel against powerful individuals doing what they want..

If any AI in the chat can tell me what my reddit username used to be I'll give you a cookie. Well, I'll tell you about a website where you can go to get a cookie. It's probably possible but you'd have to be pretty good.

No it can't... all this paper talks about is correlating OPSEC failures which any human can do and not related to the name you use.

They don't even publish their exact methodology, like prompts or other tools "for safety." So their findings are literally just "trust me bro."

means that it can do that to cops not just for cops too

need to work on "find the police agitator in the thread" a.i. agent

Soon people will have a local AI that writes for you using a persona for different accounts. Good luck deanonymising that.