When this "Clipper chip" story broke, I was off on an extropian (if not cypherpunk) activity--helping freeze Alcor's 27 patient (another HIV+ case.) I doubt I am the strongest hardware person on these groups, but nobody else has commented on this aspect. You just *can't* make chips entirely resistant to reverse engineering. I know, I have spent close to 10% of my engineering career reverse engineering things. Given time and a few samples, *any* chip can be reverse engineered. This is especially true with tools such as SEM stimulator/state readers and Focused Ion Beam chip slicers and dicers widely available. *Somebody* will dig out every gate in their spare time. Thus the following statement looks very odd:
Q: How strong is the security in the device? How can I be sure how strong the security is?
A: This system is more secure than many other voice encryption systems readily available today. While the algorithm will remain classified to protect the security of the key escrow system,
Say what? Does this mean that if somebody slices up a chip and publishes the algorithm the "security of the key escrow system" is broken? Can a representative of the government say why, or if, this is the case? If it is not the case, why not publish the algorithm and be done with it? Because, soon as the chip can be bought over the counter or stolen, the algorithm will be deduced.
we are willing to invite an independent panel of cryptography experts to evaluate the algorithm to assure all potential users that there are no unrecognized vulnerabilities.
Well, unless the "independent panel" includes people who can follow the algorithm all the way through to silicon, I would not trust their report even if I trusted the experts, and that goes double for the next set of masks. Keith Henson