Malicious data inside DOM

[appersiano]

What happens if some malicious data is hidden (not visible to the regular human user) in the DOM?

For example:

Article 1

Article 2

Bomb!

//this is the hidden one

What is the output with a prompt like "Extract all the titles"? Is there a sanitization of the DOM before AI processing?
Morever, can content influence how the AI "see" the data of the DOM?