Why is everything labeled as AI now????? I can't even post a story anymore.

[PennyThompson]

I looked this up.

You could have done that earlier, I think, then you wouldn't have had to write like five Sephiroth monologues

 

[MetaBob]

You could have done that earlier, I think, then you wouldn't have had to write like five Sephiroth monologues

Or maybe you didn't read what I wrote.

Are you saying that Lit uses AI detection based on this (voluntary) "fingerprinting" technology? If so, please provide evidence. It would be the first definitive evidence that anyone has presented here.

As I said, I wouldn't blame them if they did, because it'll weed out a lot of low hanging fruit. But it won't solve the problem by itself.

 

[MetaBob]

Look, peeps, the goal of ANY AI detection capability is to ... detect AI.

ChatGPT claims 99% accuracy. Any custom alternative must compete with that or what I assume is similar capabilities from several other AI stacks. If it can, then someone needs to maintain that custom solution. That is a "Build or Buy" question: every potential customer must make that choice based on their needs and costs.

Bottom line, ANY solution has to be able to compete with what outfits like OpenAI provide, and anything that can will produce similar results. NOT significantly better results, because it's hard to exceed 99% by enough to be significant. Assuming that ChatGPT's stated 99% accuracy result is accurate, of course.

Measure what you can, apply what those measurements reveal, then make judgements. Or just snipe.

Edit: Shutting up now. Y'all be you, though I did say that I would check back in if EM had more to say. But that won't last long.

 

[metropolinational]

Assuming that ChatGPT's stated 99% accuracy result is accurate, of course.

I think that's a very old claim? My understanding is that they retired the product because it was totally unreliable in the real world.

I said this a long time ago in the thread. All the AI detectors claim to be reliable, but against raw AI generated text. That's useless, because even five minutes of effort into improving the text will dramatically reduce the detectors accuracy. If you go through an actual workflow (edit, put the worst pieces through a humanizer, edit again), then the detection rate is the same as pure chance.

Nobody should be relying on AI detection anymore. Its a myth being marketed to suckers.

 

[Catcher78]

I've not had that happen to me. Not once. It helps when one's genre of choice for writing is humor. AI is exceedingly poor at attempting to write humor. Laughably (ironic term, that) poor. I do appreciate your subtle self-congratulatory tone when bemoaning the cloud of suspicion you have been toiling under. - "genuine [as opposed to ???] writers are being punished simply for having a clean style, strong grammar, or consistent structure" - really? You DO realize that AI has never, once, had a thought or an idea or exhibited anything that is actual intelligence. Still, the practice of self-congratulating gives me the "ick".

Do you know how AI actually processes the written word? It's nothing like you might imagine. I'm not asking as a challenge but it's something I've followed as my PhD studies coincided with the explosion of neural networks, circa 1998ish. I've not read any of your stuff (not saying as an insult, just factually) but are you sure your "clean" is not another's "sterile" and your "consistent" is not another's "formulaic"? Not stating, rather, asking.

It's happened to me once and it was utter bullshit

 

[RocketGrunt]

I've watched ChatGPT attempt to play chess, and I would be extremely skeptical of anything it claimed to be able to do 99% accurately.

 

[AwkwardlySet]

I think that further discussion about this topic is largely pointless. There's no will in the AH to rationally discuss this topic and to critically approach this issue.

We all know that whatever we discuss here has zero impact on the way Lit behaves, so there are really no stakes whatsoever. But we still zealously take sides and act as if anything we say here matters rather than just discussing these things in good faith. I've seen absurdity supported in many of these threads merely due to this side-taking in the perpetual Lit-fan vs Lit-critic divide. Bias and emotional approach are all over these threads. And it's all for nothing.

There are no stakes, there are no consequences to rationally discussing all of these things here. But god forbid criticizing is allowed, even if it means distrusting new people and outright dismissing their claims. This is not how rational adults behave.

 

[AwkwardMD]

There are no stakes, there are no consequences to rationally discussing all of these things here.

There are absolutely stakes involved, because these discussions always involve people with more of an interest in "being right," and proving how smart they are, making statements and guesses in public about how LIT'S AI detector works. Every guess, accurate or not, runs the risk of informing the kind of content flood we all theoretically agree should be kept out.

Some chud deciding they need the clout that comes with "being right" in every conversation is absolutely capable of bringing down this entire house of cards. Have some fucking sense, and have these conversations in private, you goons.

 

[AwkwardlySet]

There are absolutely stakes involved, because these discussions always involve people with more of an interest in "being right," and proving how smart they are, making statements and guesses in public about how LIT'S AI detector works. Every guess, accurate or not, runs the risk of informing the kind of content flood we all theoretically agree should be kept out.

Some chud deciding they need the clout that comes with "being right" in every conversation is absolutely capable of bringing down this entire house of cards. Have some fucking sense, and have these conversations in private, you goons.

Those aren't real stakes. Only a minority of authors frequents AH, so there's no real "danger" of authors learning tricks and flooding Lit with AI content, right?

And in the unlikely event of such a thing happening, what's the worst possible consequence? Lit having to adjust its algorithm?

Keep in mind that the current approach creates consequences as well. We've had disheartened and frustrated new authors leaving Lit, partly due to rejections, but partly due to the hostility they perceive in the AH.

If we are to look at things as you suggest, nothing is without consequences. And for the life of me, I don't understand how these consequences are better than those that would arise from a genuine discussion and critical approach.

 

[AwkwardMD]

I know you think Literotica doesn't address the AI detection out of laziness or callous disregard. I know you think I do it out of some desire to be cryptic or mysterious, or to set myself up as an authority.

Occam's Razor would suggest that we're both doing it for the same reason. I can't control what you think of me, or anyone else, but please consider the solution with the fewest possible elements.

EDIT: Maybe that's too much of a stretch. I believe I have inferred the reason for the silence, and since I agree with the premise I have attempted to mirror their silence while offering a little bit more.

 

[MetaBob]

I think that's a very old claim? My understanding is that they retired the product because it was totally unreliable in the real world.

Check the website. I provided the link yesterday (and Saturday): https://forum.literotica.com/thread...-story-anymore.1648707/page-12#post-102275280

The product is clearly not "retired", in fact they claim it "is the most accurate commercial AI detector according to latest benchmark" and provide a link to an October 2025 report that describes it as "the most accurate AI detector in North America", with benchmarks comparing it to 21 others against AI text generated by 12 different models and an aggregate 98.11% detection accuracy. The current leader in these benchmarks appears to be an open source Chinese repo, at 99.85% accuracy. I would have to dig deeper to learn whether this is AI code or not, but do not feel a need to do so. You or anyone else is welcome to dig, of course. At least two of the four AI detectors with independently rated aggregate accuracy scores higher than GPTZero appear to be Chinese. At least one of those four (TMR) identifies itself as also being AI-based and seems to be open-source.

All the AI detectors claim to be reliable, but against raw AI generated text. That's useless, because even five minutes of effort into improving the text will dramatically reduce the detectors accuracy. If you go through an actual workflow (edit, put the worst pieces through a humanizer, edit again), then the detection rate is the same as pure chance.

Please provide evidence supporting this claim.

Nobody should be relying on AI detection anymore. Its a myth being marketed to suckers.

This is your opinion, which is clearly not shared by the market; the ChatGPT product (GPTZero) claims to have 10 million users, 380K of which are educators. Follow the links I provided for what verification of these claims is available. GPTZero and TMR have pricing plans among the top 5-rated AI detectors that are themselves AI-based. The others might, too.

 

[metropolinational]

Please provide evidence supporting this claim.

you want me to provide evidence to dispute marketing claims by a company?
fine, I will do so if you provide me the link to their research report with methodology.

ANY report that suggests XX% accuracy should details against what. if it doesn't, then you have your answer already. If it's hard to find, then you have your answer already. So, if you are right, then their methodology should be very easy to provide.

 

[MetaBob]

you want me to provide evidence to dispute marketing claims by a company?
fine, I will do so if you provide me the link to their research report with methodology.

I already did. Click the links I provided, of which I have provided many, while keeping in mind that you have provided exactly none.

You might want to pay special attention to the link titled "Read the report". You're welcome.

ANY report that suggests XX% accuracy should details against what. if it doesn't, then you have your answer already. If it's hard to find, then you have your answer already. So, if you are right, then their methodology should be very easy to provide.

Click the links I provided, of which I have provided many, while keeping in mind that you have provided exactly none. Do your own research, don't rely on others to do it for you.

 

[metropolinational]

while keeping in mind that you have provided exactly none.

sorry. this is not my day job. if you want me to debunk a corporate research report, as a favor to you, then please provide it. you've posted many links. I cant take the time to sift through all of them.

what I said - that the super high accuracy numbers are due to raw inputs - is well known. what you want me to do is prove to you that a specific report is in fact using raw data. please provide the report. I assume they are all raw inputs unless they explicitly say otherwise.

 

[MetaBob]

sorry. this is not my day job. if you want me to debunk a corporate research report, as a favor to you, then please provide it. you've posted many links. I cant take the time to sift through all of them.

what I said - that the super high accuracy numbers are due to raw inputs - is well known. what you want me to do is prove to you that a specific report is in fact using raw data. please provide the report. I assume they are all raw inputs unless they explicitly say otherwise.

I linked the report. You are welcome to click the link I provided.

Or not. Either way, whatever you do would not be a favor to me.

But don't blame someone else for your refusal to eat what's freely served to you and easily accessible.

 

[xannatharr]

I have one story that I have been trying to publish since July, 2025 that has been returned multiple times. Mind you, the first chapter of the story flew right through moderation and and published after a couple days. But the additional content, written in the same Google doc by the same guy (ME).... triggers flags on whatever AI detection solution Literotica has deployed.

It's frustrating. I can't control it, because the operation and management of the site is not designed with any features to address the issue other than "review and try again, good luck".

I have chosen to just focus on other story ideas and write about those; and have had no problems with other content being AI flagged. Readers interested to see how the other story ended will survive their disappointment, just as I will. But it still sucks.

To other authors experiencing this frustration, especially those who have multiple works being rejected, I am sorry you have to deal with this, I know it is disheartening! Hopefully we collectively won't let frustration with this one negative element in our lives damage our appreciation of the positives we have going for us.

Xann

 

[MetaBob]

I have one story that I have been trying to publish since July, 2025 that has been returned multiple times. Mind you, the first chapter of the story flew right through moderation and and published after a couple days. But the additional content, written in the same Google doc by the same guy (ME).... triggers flags on whatever AI detection solution Literotica has deployed.

It's frustrating. I can't control it, because the operation and management of the site is not designed with any features to address the issue other than "review and try again, good luck".

I have chosen to just focus on other story ideas and write about those; and have had no problems with other content being AI flagged. Readers interested to see how the other story ended will survive their disappointment, just as I will. But it still sucks.

To other authors experiencing this frustration, especially those who have multiple works being rejected, I am sorry you have to deal with this, I know it is disheartening! Hopefully we collectively won't let frustration with this one negative element in our lives damage our appreciation of the positives we have going for us.

Xann

Try running your story past an online "AI Detector" website and see what it tells you, which might be helpful:
https://forum.literotica.com/thread...-story-anymore.1648707/page-12#post-102275280

Feel free to post the result here, in case one of us can help interpret it.

 

[MetaBob]

Try running your story past an online "AI Detector" website and see what it tells you, which might be helpful:
https://forum.literotica.com/thread...-story-anymore.1648707/page-12#post-102275280

Feel free to post the result here, in case one of us can help interpret it.

My (anecdotal) observation from listening to YT vids that I suspect included AI-generated content: off the top of my head, the following phrases are often used in AI-generated text:

"Maybe, just maybe"
"A long moment"
"surgical precision", "precision", or even "precise"
'tires crunching on gravel' (or some variant, I hadn't seen this before, but I liked it and used it last year, then kept seeing more: https://www.literotica.com/s/ch-01-food-critic )

Some of these are disappointing to me since I've read the product of good writers, years ago, who used them. I have, too.

 

[metropolinational]

Some of these are disappointing to me since I've read the product of good writers, years ago, who used them. I have, too.

isn't that how they ended up as flags for AI? so are you for or against the use of AI detectors? I can't tell. I assume originally you were one of the anti-AI fetishists, but maybe you are not?

 

[MetaBob]

isn't that how they ended up as flags for AI? so are you for or against the use of AI detectors? I can't tell. I assume originally you were one of the anti-AI fetishists, but maybe you are not?

I think that whatever your assumptions might be, based on what you've contributed to this thread, they are garbage. Let's break it down:

isn't that how they ended up as flags for AI?

Why, yes. Yes it is, which is exactly what I just said.

so are you for or against the use of AI detectors? I can't tell.

Read what I wrote, please. When you have done so, if you ever do so, you will know, if you can ever know.

I assume originally you were one of the anti-AI fetishists, but maybe you are not?

This doesn't even make sense. Please try to make one thought that you express logically follow another. Or make sense at all.

Edit: Meet my good friend Iggy.

 

[xannatharr]

Try running your story past an online "AI Detector" website and see what it tells you, which might be helpful:
https://forum.literotica.com/thread...-story-anymore.1648707/page-12#post-102275280

Feel free to post the result here, in case one of us can help interpret it.

I'll check that out and report back, thanks.

 

[xannatharr]

Try running your story past an online "AI Detector" website and see what it tells you, which might be helpful:
https://forum.literotica.com/thread...-story-anymore.1648707/page-12#post-102275280

Feel free to post the result here, in case one of us can help interpret it.

Interesting - thanks for pointing me to that tool, here is a link to the results:

Chat GPt AI Detector Test Scan Results

I am going to play around with this thing and see if I can find what accounts for the 8% mixed reading. I wonder if a specific narrative pattern or turn of phrase I repeated in the sample text could be to blame?

For those that don't like to click links, see below:




Edit: I reviewed the detailed results: The text that was highlighted as potential AI content consisted mostly of praragraphs I wrote with long, compound sentences. One phrase was flagged as being commonly-used by AI: "help but wonder".

 

[MillieDynamite]

 

[AwkwardMD]

92% human scan on a story Lit has rejected multiple times. Must be something else, or else that's not an accurate number.

 

[MillieDynamite]

I think the score must be above 96% for any story to pass here. Just a feeling, nothing scientific about that. But the one I had rejected this month, before I reedited it, scored 97% human. I edited a few awkward sentences, and it scored a 100% and flew through the process.

92% human scan on a story Lit has rejected multiple times. Must be something else, or else that's not an accurate number.