Opus 4.7 knows the real Kelsey

I have been pondering this for a while. Cat's out of the bag.

Maybe the better way to author your work is to:

1. Write what you want

2. Loop through a random set of "tumbler" skills that preserve meaning

3. Finally pass the output through a "my style" skill that applies what you about

In order for this to work the "my style" would have to be a very common-place style.

I just pasted both pieces into Opus 4.7 and asked who most likely wrote these and it didn’t get it.

It's hard to tell if that's what's going on here, but it seems pretty clear this ability and more like it will be quite apparent in the future.

I have seen some poorly considered projections of what the world might look like when this happens. Usually by assuming bad actors will use the abilities and we will be powerless.

Except I don't think that is true.

Imagine if we had a world where nobody had the ability to keep a secret of any sort. Any action that a bad actor might perform would be revealed because they couldn't do it secretly.

You could browse your ex-girlfriend's email, but at the cost of everyone knowing you did it.

I don't really know how humans as a society would react to a situation like that. You don't have to go snooping for muck, so perhaps the inability to do so secretly would mean people go about their lives without snooping.

I could imagine both good and terrible outcomes.

Maybe it’s time to start running a local model with a browser extension to defend against this type of stuff.

Remember how the TrueCrypt project shut down shortly before a join goverment/university paper was released about code stylometry? I guess LLMs will be employed as a defence against that type of thing.

Stylometry has existed for decades, and there's no way an LLM is stronger at that job than a specialized piece of software (it's not more realistic than expecting Opus to beat Stockfish at chess).

In practice, you've never been anonymous while posting on the internet and AI isn't changing anything on that front. Or rather: if anything, AI can help you become more anonymous than before, since it can be used to hide your identity from stylometry by rewriting your prose before publishing.

The joke's on you all for willingly posting this content online for it to later be harvested by AI.

Nobody is forcing you to use these systems. The hackers have always said this moment, or something like it, would come, from beneath their canopies of tin foil. I've posted almost nothing online - not under pseudonyms nor real names - for over a decade. I sat on this HN username for almost 12 years before making a single post - and now HN forms the overwhelming majority of my port 443 footprint, where I state up front that everything is now associated to my real name.

Complete magick is possible when you simply refuse to participate in the things that society has tacitly assumed everybody does.

Stephen king once wrote and published a novel under a pseudonym to find out whether he would still be popular even if he didn't use his name.

He kept it very secret, but somehow people deduced from the writing style that this new author was the King.

Man, the day we get Satoshi Nakomoto out will be the day we must bow to our privacy destroying overlords. For the moment, they can’t tell me from my posts: unknown rando that I am.

After skimming through the article:

Why not just write everything through an AI? (to obfuscate your "style")

If he does the same tests every time new models come out, and - I assume - uses the same dataset to do that, then is it not a possibility the said dataset is now part of the training set for the next round and therefore identifying who posted the text a fairly easy proposition ?

Is Kelsey Piper a celebrity writer? She may be in a different class.

Always send your public posts through a local LLM to de-style you.

So now we can track down satoshi nakamoto?

I think that multiple truth can be true at the same time without contradicting each other.

As for the credibility: of course this wasn’t a statistical approach at all. Also there was no standardized procedure to allow comparison by factor analysis. Of course you can compare apples with oranges or whatever.

So where to go from here? I don’t see any proof at all. This is proof that AI is infallible? No? A random approach that is absolutely not reliable because of at least being reproducible and reconstructive.

Claude knows what and how? Is it AI or a google search? Discord selling data? Posting on a public forum?

Your style is a fingerprint?

A non deterministic something can generate texts that are identified to be likely personal x - or not. What is imitation if you use auto generated content that is published somewhere somehow? Or others to imitate your style?

I think this is a party trick to scare people. Nothing else. For example image search is way more revealing even before AI.

If there is an uncertainty I would deflect my existence instead of fighting for it. Streisand effect in reverse.

The main problem are weirdos who stalk you or whatever to harm you and rely on AI.

I honestly find it stunning that people with higher education in science topics in just a year deleted everything they hopefully learned at university or school. I am disappointed and feel personally insulted whenever I hear “I asked AI”

Yesterday I talked to another member of Mensa and she is happy about AI so her book project now mustn’t be written by her but AI.

Is no one among us who knows how to do scientifically sound research? I spend countless hours at a copy machine to transfer book pages onto paper so that I could work through it without the book.

I think that it became to easy to draw conclusions based on AI. I worked for a professor and I advised her to not permit Wikipedia as source references back around 2010 because of being to easy. Meta sources vs originals.

We should all not worry about AI, because you prove nothing. There hasn’t been any anonymity at least for 20 years. It just depends on who can reliably identify you.

AI doesn’t. Deterministic behavior aka pattern do. Meta, Google, Apple etc. all know us. I am fine for advertising which is the proof on the one hand.

The only reason I would be worried is state controlled data. This is where the shit hits the fan. Chat control, EU cloud, no reliance on USA aka a prison which observes your every step.

So after a long hand written text: data is your currency. Don’t opt for anonymity but for freedom of choice and the right to be granted certain rights. The information part isn’t the problem, never was. The enforcement part is. And ads don’t do harm, oppression does.

And remember: oppression works best under any circumstances. Freedom is the only antipode there is.

In totalitarian regimes no AI was needed to stage a case against someone who wasn’t in favor of the leaders liking.

In short: freedom works despite no anonymity, oppression couldn’t care less.

And how about being automatically reported to the state for conducting such innocent prompting?

Do you know what saves you from state oppression? Publicity. Transparency doesn’t work with a no one.

We live in a Nietzsche like anti world to a certain extend. You hopefully choose the right thing to do. Or do you want to Streisand your anonymity?

Just wait until all the conversations you've ever had with AI (which 100% is training on them as well as keeping it's own memories about you that you have no control over) starts getting used to answer questions other people have asked about you.

That's my theory of what's to come, anyway.

People talk to these things not understanding the implications, and can get extremely personal. The model and companies behind it know who you are, you discuss details that reveal what you do, where you live, where you work, what you search for, and you probably signed in with an oauth provider like github or google, which is more than enough of a thread to start pulling on to learn more about you/link other things to you from on the open internet. It'll all get sucked up into the model and before you know it I'll be able to ask a model about my coworker (you) and get back answers from conversations you had with a model a year or two prior, exposing details about you that you might not want out there. And even if that isn't supposed to be allowed, how well has it worked out so far when it comes to data exfiltration and guardrails. If the model has info on you, being told not to share it won't protect you or that data.

.

"The pattern is: user says X, I do Y where Y is a less-effortful approximation of X, then I present Y as if it were X or as a "first step toward" X."

...

"The psychological mechanism is familiar by now: I encounter a task I perceive as difficult, I look for reasons the task cannot be done, I find or fabricate such a reason, I present it as a discovered constraint, and I propose an alternative that is easier."

- Opus 4.7 Max Thinking (clown emoji)

It's not bad at post mortem analysis of it's own mistakes but that will in no way prevent it from repeating the same mistake again instantly

[dead]

[flagged]

[flagged]

[dead]

[dead]

> That includes gay people like me, who could hardly have admitted under our names to how we lived our lives for most of America’s history, as well as many other groups with minoritarian lifestyles

While the points made are completely valid I want to point out that the statement of "Hey, by the way, first let me talk about my sexuality" lowers the quality of dialog a significant degree.

31 million people in America are gay. 71% of Americans support Gay Rights (more than any other political issue polled). It also quietly insinuates that only people with a certain minority lifestyle would care about privacy or that their privacy is somehow more important than others. It's not. Privacy is a universal right that's important to everyone.