It looks exactly the same as a standard LLM-generated project to me with all the hallmarks of it. The titling with all small caps with spacing, the font choice, the inlined SVGs that are not quite there nor identical to a modern library, the default of trying to use characters or emojis or unicode special characters instead of SVGs for the correct/incorrect in help.
A human example: A designer might have checked breakpoints for mobile and emulated the mobile chrome but if the user is actually on mobile dropped the fake phone chrome and re-distanced the icons so that it would simply be a normal page.
It still has some of the AI slop marks though. The background gradient for example.
Overall it's not consistent and seems to mix different styles which is different from the typical AI slop that is consistently bland and generic. It does look off and eerie to me.