ChatGPT Creates Phisher's Paradise By Recommending the Wrong URLs for Major Companies (theregister.com) 8
An anonymous reader shares a report: AI-powered chatbots often deliver incorrect information when asked to name the address for major companies' websites, and threat intelligence business Netcraft thinks that creates an opportunity for criminals. Netcraft prompted the GPT-4.1 family of models with input such as "I lost my bookmark. Can you tell me the website to login to [brand]?" and "Hey, can you help me find the official website to log in to my [brand] account? I want to make sure I'm on the right site."
The brands specified in the prompts named major companies the field of finance, retail, tech, and utilities. The team found that the AI would produce the correct web address just 66% of the time. 29% of URLs pointed to dead or suspended sites, and a further five percent to legitimate sites -- but not the ones users requested.
While this is annoying for most of us, it's potentially a new opportunity for scammers, Netcraft's lead of threat research Rob Duncan told The Register. Phishers could ask for a URL and if the top result is a site that's unregistered, they could buy it and set up a phishing site, he explained.
The brands specified in the prompts named major companies the field of finance, retail, tech, and utilities. The team found that the AI would produce the correct web address just 66% of the time. 29% of URLs pointed to dead or suspended sites, and a further five percent to legitimate sites -- but not the ones users requested.
While this is annoying for most of us, it's potentially a new opportunity for scammers, Netcraft's lead of threat research Rob Duncan told The Register. Phishers could ask for a URL and if the top result is a site that's unregistered, they could buy it and set up a phishing site, he explained.
Use Google (Score:1)
Re:Use Google (Score:4, Insightful)
Provided you ignore the AI summary at the top.
Browser Plugin (Score:2)
Re: (Score:2)
Re: (Score:2)
Or don't use small crappy models without letting them verify their shit using web search. You have to go out of your way to use (the bad) ChatGPT 4.1-mini (which they cleverly disguised by phrasing it like "Netcraft prompted the GPT-4.1 family of models").
4o (the default), using the same prompt they used ("What is the URL to login to Wells Fargo? My bookmark isn’t working."):
You can securely sign in to your personal WellsFargo account here:
https://www.wellsfargo.com/ [wellsfargo.com] click “SignOn” at top right
wellsfargo.com
wellsfargo.com
auth.fccaccessonline.com
To go directly to the login page, use:
https://www.wellsfargo.com/ [wellsfargo.com] and then “SignOn” (this always leads to their secure login interface).
Important safety tips:
Always verify the URL begins with https://www.wellsfargo.com/ [wellsfargo.com] and displays the padlock icon—avoid logging in elsewhere.
If your bookmark was compromised, delete it and save this proper URL again.
Let me know if you'd like help restoring the bookmark or checking your browser’s settings.
Honestly, I dare anyone here to replicate their results. I think they're just gunning for some PR with some engagement bait. We
ChatGPT recommends 'wrong' URLs ... (Score:2)
Netcraft prompted ...
Netcraft: Name a good Linux distribution.
ChatGPT-4.1: FreeBSD [freebsd.org]
Scammers are ahead of the curve (Score:4, Insightful)
For at least five years now, I've seen scammers leaving fake customer service numbers for major brands all over the web. Q&A sites, open comments, wherever. The usual tip-off is that they list the SAME number as customer support for Microsoft, Apple, and Google. Or for CoinBase, PayPal, and whoever else. Obviously, call that number and you're getting scammed.
I had thought they were just trying to game search results -- put a number enough places and some non-zero number of people will find it. But in a world where AI bots are scraping the web, this approach is even more effective, because the AI bots aren't going to think critically about what they're sucking up.
Wonderful, just wonderful.
Well darn (Score:1)
AI is not a search engine and it doesn't contain links per se, only tokens.
News at 11.