Usually copyright laws refer to a substantial part of a work, it maybe be 0.6% of the delivered web page, but it's 100% (presumably) of the article. The article counts separately as a work (and each image within an article could count as a separate work, there are conditions on that). But, in any case if all the textual context was copied that's still substantial. Interpreting things in a purely numerical way like that would be clearly ridiculous.

Such gotchas don't work in law in general (in my very limited experience).