gives the 40 million (35+5). It gives by name a bunch of genes unique to humans, though doesn't count them. 300 is just an estimate, I couldn't find a specific reference to that number, thought it's widely to believed to be in the hundreds (and not, say, the millions).
There may only be a smallish number of genes that are unique to humans or chimps, but obviously that is not the only source of differences between the two species. The first press release you linked states that,
> At the protein level, 29 percent of genes code for the same amino sequences in chimpanzees and humans.
There are about 20,000 known protein-coding genes in the human genome [1], so that alone refutes the notion that there are only about 700 differences unique to humans. Besides novel unique and changes in the protein sequences themselves, changes in gene regulation is another obvious source of differences.
See this for a more thorough, and up to date, review
