[mpeg]

I don't mean to be overly critical of your product, as I hate when people do that on HN

However, you should know blurring is not a secure way of censoring information, as each character is still sort of recognisable, and programmatically reversing the blurring of known alphabets and typefaces is not that difficult (for numbers, you could blur each number from 0-9 and look at the resulting images)

This is even more obvious in the text that includes caps and descender characters (q/g)

Maybe you should consider a black rectangle instead?

[Tepix]

I agree, the blurring is not "strong" enough, too much information is retained. It's highly likely that it can be undone (I mean if blurred stuff gets passed around as a screenshot - it would be trivial to undo it right on the webpage ;-) )