- Chatgpt cannot know if a site was pirated, expired or reused for the casino spam
- The answers generated by AI may seem reliable, even when they cite completely kidnapped and false sources
- The defeated charity domains are reborn as game sites and still pass as reliable sources
Chatgpt is quickly becoming a reference source for people looking for recommendations, from online services to local companies, but the new evidence suggests that their suggestions generated by AI may not always be based on reliable sources.
In fact, some are extracted from websites that have been pirated or whose domains have expired and reused, often to promote online casinos and game platforms.
In recent months, James Brockbank, managing director and founder of Digitaloft, has been documenting how extended the problem has become, discovering examples of chatgpt citing sites content that has clearly been manipulated.
Exploitation of gaps in the validation of the source of AI
In one case, the website of a functional legal practice, directed by the lawyer Veronica T. Barton, had pages that recommended casinos of the United Kingdom buried within it.
“Its site has been pirate and this page added,” BrockBank said after reviewing the evidence.
In another case, a site once affiliated with a United Nations youth coalition had become a platform that pushed “casinos not in Gamstop.”
Although the list that housed contained only one external link, it led to another reused domain.
The pattern continued with expired domains, including one that had belonged to an artistic beneficial organization now previously linked by the BBC, CNN and Bloomberg.
The chatgpt cited by the domain, now pressing the content of the game, in response to a consultation on casinos without deposit.
These tactics exploit the weaknesses on how Chatgpt selects and cites sources, as unlike traditional search engines, the model lacks mechanisms to verify the legitimacy of the property or editorial intention of a site.
As a result, the content injected into compromised websites can arise in their answers without any obvious red flag to the user.
Chatgpt seems to favor the recent content and still attribute authority based on the reputation of the inherited domain, even when the contents of the domain has no continuity with its past, which opens the door to the bad actors to manipulate visibility through means that have little to do with credibility.
The conclusion is that users who resort to ChatGPT to obtain recommendations should not assume that each response is backed by a credible source.
A rapid verification of the authority of the cited site, its history, property and relevance can be very useful to avoid deceptive or harmful suggestions.
