It is defined by RFC 3987.
There are reasons to see URIs displayed in different languages; mostly it makes it easier on users who are unfamiliar with the roman alphabet, and assuming that isn't too difficult for anyone to replicate arbitrary unicode on their keyboards this can make the URI system more worldly and accessible.
Mixing IRIs and ascii URIs can make it much easier to do phishing attacks which trick someone into believing they are on a site they really are not on. For example, one can replace the "a" in www.ebay.com or www.paypal.com with an internationalized look-alike "a" character, and point that IRI to a malicious site.
Additionally, it can be difficult for those with different language keyboards to access web resources in other languages; in contrast, open-source programming projects (and most programs) are almost exclusively written using the Roman alphabet to avoid this type of encoding incompatibility.