[tptacek]

At most, this device seems trustworthy exactly to the extent that you trust that hardware hasn't been tampered with. Since there's very little information about how the tamper-proofing works, I tend towards skepticism.

At worst, they've implemented a cryptosystem that, having never been peer reviewed, is vulnerable to basic attacks. Since I can't find out who the authors are, and they use both AES and Serpent with listed key sizes instead of talking about how the design works, I tend towards skepticism.

Finally, I'm with everyone else wondering why deniability is something the market cares about.