GPT Countermeasures you should know

[bartosimpsonio]

Currently, AI investigators are deploying GPT countermeasures that you should probably know about.

This python script claims it can detect GPT content:

https://twitter.com/x/status/1613229743742287872

This guy launched a tool to check for AI content:

https://twitter.com/x/status/1610481575933939713

Another tool:

https://twitter.com/x/status/1613087701548535808

ALWAYS hand tweak your AI content. Always.

Same idea with AI translators. By translating original content using AI (google, amazon, etc) and then tweaking by hand, you got infinite content at your fingertips. Just requires a bit of work, but lasts a lifetime.

AI content also requires hand tweaking.

The key here is to dive into the AI content detection tools and understand what they're looking for. That's where the million dollars is.

GL!

 

[TheMarquis]

It's not about GPT itself, but about the one (the human) who is using it.

 

[Adkins]

As I wrote in another thread. They can only detect repetitions. Adjust your prompt accordingly and you're fine.

 

[nohupt]

this whole thing is getting intense lol

a real AI content generator shouldn't be detectable

 

[Adkins]

If this was that great in detecting content, they would provide a free query. They're just making money with a simple tool. Like literally anyone with basic programming language could program a "detector"

 

[TheMarquis]

As I wrote in another thread. They can only detect repetitions. Adjust your prompt accordingly and you're fine.

Not only repetitions, but also patterns. Example:
123 ?
432 ?
abcd ?

- or -

123?
432?
abcd?

-- Which one is human written? --

 

[Adkins]

Actually the first one is just wrong. But if you have a constant misspelling it is also a (human) pattern.

 

[roydan]

Very soon we'll have more AI content detectors than AI content generators.

 

[paraztos]

Very soon we'll have more AI content detectors than AI content generators.

It's time to make an AI Content Detector Detector: Is your AI Content Detector working or just feeding you BS? Find out for $100 per check.

 

[Pirsultan]

Stop words... don't nt forget ...it has a pattern that visible even for human, spelling error also can help passed detection but content with lots spelling error will make us look unprofessional

 

[Danjose]

they're simply cooking things up as they go now.

 

[bartosimpsonio]

this whole thing is getting intense lol

a real AI content generator shouldn't be detectable

Early GPT was probably undetectable.... they dumbed it down a lot in recent weeks IME

They're inserting "watermarks" now....text repetitions and so on

 

[HenryObi]

The key here is to dive into the AI content detection tools and understand what they're looking for. That's where the million dollars is.

I saw a thread somewhere, where it is reported that open AI would start including recognizable patterns which would be used to easily identify AI content. If they eventually do that. All these tools would become redundant and Google might have to rely on them.

 

[bartosimpsonio]

I saw a thread somewhere, where it is reported that open AI would start including recognizable patterns which would be used to easily identify AI content. If they eventually do that. All these tools would become redundant and Google might have to rely on them.

Yeah. Always hand edit the content, don't use verbatim AI content, it's easily detected after their latest updates. They're making it repeat some stuff over and over, it's getting obvious.

 

[Squrr]

I suspect those who copy paste directly from ChatGPT will be in for a nasty surprise one day.
I'm running a site that uses purely copy/paste so I can hopefully identify when/how it gets penalised.

 

[HenryObi]

They're making it repeat some stuff over and over, it's getting obvious.

This is interesting. If it's just repeating stuffs, say a particular word or phrase then it'd be easy to get around.

 

[Thunder Digitals]

Too many AI detectors are coming out.

 

[leit]

I'm an AI, lol.

Article written by me in 2020, ranked in first position since then. These guys want to make money by scamming people.

 

[Gebb]

View attachment 239510
I'm an AI, lol.

Article written by me in 2020, ranked in first position since then. These guys want to make money by scamming people.

I've tried as well on a 2018 article and they detected AI, it would be interesting to run their tool on their own tool's websites and see whether the output detects AI content as well. Maybe they don't detect AI content effectively but at least we can detect BS with them.

 

[BetterSEO]

View attachment 239510
I'm an AI, lol.

Article written by me in 2020, ranked in first position since then. These guys want to make money by scamming people.

Correct. I got similar results with my own written content.