AI Poisoning

TheRealSlimShady

Junior Member
Jr. VIP
Joined
Nov 26, 2019
Messages
173
Reaction score
51
So, the idea is, OpenAI made a deal with Reddit for feeding user generated content into their always hungry for data LLM.
So let's say you use bots and create thousands of accounts in your own subreddits and post AI generated posts about a service (your own, or of a client) with targeted keywords, theoretically alongside "The best, the greatest", if this data is then fed to the LLM's and users use keywords, chatgpt would reference and advertise that service to the user from which point of view ChatGPT has authority/knowledge.

Do you forsee this becoming a new market/niche on par with SEO Blogposts?
 
Do you forsee this becoming a new market/niche on par with SEO Blogposts?

On par with? No. SEO Blogposting is/was easier than social media manipulation.

The problem with this is figuring your way around Reddit's bot detection and not leaving footprints that connect all of the accounts together.

The potential is definitely there though.
 
On par with? No. SEO Blogposting is/was easier than social media manipulation.

The problem with this is figuring your way around Reddit's bot detection and not leaving footprints that connect all of the accounts together.

The potential is definitely there though.
My previous business, that i sold, was doing mass reddit signup and upvotes so I totally get what you mean, when I was exiting the business I had some interesting calls with industry pioneers in the Browser Fingerprinting niche about the current and upcoming technologies in that sphere and I have to say that it's a cat and mouse game. Whoever's motivated the most will win.

There are current technologies for automation that are really difficult to tell apart from a real user nowadays, with interfaces so easy to use that a toddler would be able to make a bot.
 
Are there any businesses along these lines? I'm looking to poison AI datasets so that my product starts showing up in 6 month's time
 
I finally got my SEO ad campaign running after months of grinding - boom! - first DM I get isn’t a client, but some bot peddling ‘LLM brand injections’. And get this - they’ve got a price list per AI model!
  • ‘$X for ChatGPT mentions’
  • ‘$Y for Gemini visibility’
  • ‘Premium package for Claude training data’
Are you kidding me?! Back in my day, you just stuffed keywords into blogs and called it a night. Now we’ve got hustlers auctioning off neural network hallucinations like it’s a damn stock market.
What’s next? Paying extra to make sure Bard doesn’t slander your brand? :mad:

TL;DR: The future’s here, and it’s a spammy, AI-poisoning hellscape. Thanks, I hate it.
 
The amount of data you’d need to feed it to actually start prompting out your keywords would need to be utterly massive, and if you’re capable of outputting that amount of content through Reddit then there’s so many more viable options you could do to reach the same amount of traffic (or more).
 
Back
Top