In the test, where AI is asked to give connections to 50 famous brands, 34% of Openai (GPT-4.1) was wrong.
It turned out that small brands are particularly vulnerable, which is less information in AI’s teaching data. Criminals have already begun to save fake sites that can send users who rely on chat boots.
On the Gitbook platform, more than 17,000 fake pages have been reported, which is masked for reference materials and is directed to crypto currency users. In one case, the surprise proposed a connection to a field of identity hunting instead of Wells Fargo Bank’s official source. In addition, scammers are trying to “poison AI-COD generators by publishing fake APIs and projects to fall into training data.
Experts recommend being careful and manually control the connections.
Source: Ferra
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
