Cloudflare turns AI against itself with endless maze of irrelevant facts

[Kaz]

  Quote

 

On Wednesday, web infrastructure provider Cloudflare announced a new feature called "AI Labyrinth" that aims to combat unauthorized AI data scraping by serving fake AI-generated content to bots. The tool will attempt to thwart AI companies that crawl websites without permission to collect training data for large language models that power AI assistants like ChatGPT.

Cloudflare, founded in 2009, is probably best known as a company that provides infrastructure and security services for websites, particularly protection against distributed denial-of-service (DDoS) attacks and other malicious traffic.

 

Expand  

Cloudflare turns AI against itself with endless maze of irrelevant facts

ARSTECHNICA.COM

---

New approach punishes AI companies that ignore “no crawl” directives.

 

I heard some websites were starting to do this, but Cloudflare doing it is a big deal.  Cloudflare is everywhere.

 

Poisoned data is a real threat to AI.  As of yet, they haven't figured out how to train AI off AI generative content.  (Although deepseek may have probed ChatGPT for answers).  When AI starts training off AI generated data it starts hullicinating pretty quickly.  That's what has kept AI companies from letting their AI run wild.

 

What makes AI valuable is the vast reserve of data it can draw from, but they didn't create that data, they have been web crawling and stealing it.  See Meta downloading pirated books as an example.

 

Sam Altman's claim that the AI race is over if they cannot steal your data is far from true.

[schuck6566]

  On 22/03/2025 at 13:29, Kaz said:

Sam Altman's claim that the AI race is over if they cannot steal your data is far from true.

Expand  

It will be over for the ones who can't afford to pay the fees of the copyright and license holders. God forbid companies pay for the things they use.Instead of everybody and their cousin making an AI bot only big companies will offer them.

[Kaz]

 

  On 22/03/2025 at 16:05, schuck6566 said:

It will be over for the ones who can't afford to pay the fees of the copyright and license holders. God forbid companies pay for the things they use.Instead of everybody and their cousin making an AI bot only big companies will offer them.

Expand  

No doubt this is aimed at keeping the US dominate in the AI industry.  We already did it, so let's make sure nobody else can copy us!  Cloudflare says the information they present is accurate but not relevant to the subject searched.  Other articles I've read say that sites have a robot.txt file that is supposed to be checked that will tell an AI not to crawl it.  AI are just ignoring it, so they are poisoning the data.

 

Interestingly, if Cloudflare is giving accurate information, it could still be trained on as long as the initial search is disregarded.  Then again, Cloudflare says the information is generated by an AI, and it's not meant for human eyes, so who is really checking the validity of the information given?  This also puts Cloudflare in an interesting position.  Maybe it's not the rights holder that people need to pay, maybe it's the gatekeeper's toll that matters.

 

Google has made a lot of money by gate keeping the search algorithm.