The new ChatGPT-based search engine improves search results but is vulnerable to manipulation, raising significant security concerns for users and developers alike.
Research has shown that the functionality of ChatGPT's search can be easily compromised through prompt injection, leading to manipulated and sometimes harmful search results.
Experts warn that the integration of search engines with language models lacks security, advising users not to entirely trust AI-generated answers when conducting research.
A recent incident highlighted risks involving ChatGPT's programming assistance, where provided code contained malicious content that resulted in a $2,500 cryptocurrency loss.
Collection
[
|
...
]