They could get around this via an open source browser plugin so that it can be peer-reviewed, or (more desirably) the functionality can be integrated into browsers via the HTML5 spec.
There's no such thing as an Appleseed binary, btw, it's written in a scripted language.
BTW, I'm not trying to imply that these guys have solved the problem -- there are numerous problems, as you pointed out.
Have you read the link I posted about how you could approach encryption at the browser level?
Far from not solving the situation, I think they fall into the category of "half-way encryption is worse than no encryption at all", and they don't seem to understand the social concept of trust. As Jon Callas says, "Encryption is not magic pixie dust that makes everything okay."