The thing is called "tracing" and quite common.
The actual code (which means the ttf/otf) is protected under copyright. The shape of a letter isn't and can't be by any form of IP.
So: You're not allowed to redistribute the font file but creating your own font which just happens to look alike is perfectly fine.
So, a given "specification" (how a font looks to the human eye) can have many different "implementations" (ie. TTF files). The specific "implementations" themselves can be copyrighted, but somehow not the "specifications"?
If so... assume one TTF file which is copyrighted, pay-per-use license. Assume another TTF file, which is open-source, free to use and redistribute. The files both implement the "Helvetica" specification, rendering the same letters to the human eye. How "different" must the free implementation be from the paid one, for it to not be considered an infringement?
They probably won't however, since you might be able to copy the shape, but for being displayed on screen you need a process called "hinting". That is the term for making pixels out of the shapes. Professional fonts are hinted by hand, making it look neat even in small sizes. There is auto-hinting, but it only gets you so far.