Re:Surreptitiously Tampering with Computer Chips
by makomk on Monday September 16, 2013 @03:08PM
(#6395145)
I remember seeing the original journal article a while ago on Slashdot (somehow it never made it to the front page on HN). It's really ingenious - even inspection under a microscope can't spot the difference between the tampered hardware RNG and a honest implementation, it passes all the tests, and yet it actually has so little entropy that essentially anyone can break otherwise-secure crypto that uses it.
[ Parent
| on HN ]
Re:Surreptitiously Tampering with Computer Chips
by pavel_lishin on Monday September 16, 2013 @03:32PM
(#6395316)
> yet it actually has so little entropy that essentially anyone can break otherwise-secure crypto that uses it.
Isn't this a testable thing?
Re:Surreptitiously Tampering with Computer Chips
by makomk on Monday September 16, 2013 @03:35PM
(#6395349)
Possibly, if you know what you're doing, but as someone in the comments at Bruce Schneier's blog points out the changes are small enough that it's feasible to only backdoor selected batches of chips (for example, just the ones shipped to countries of interest to the US or companies like Google).
Re:Surreptitiously Tampering with Computer Chips
by tlb on Monday September 16, 2013 @04:02PM
(#6395527)
In fact, it can also be designed as a marginal circuit so that a small fraction of the chips fail due to manufacturing variance. You can test for the weakness and ship the bad chips to people whose privacy you don't respect.
Re:Surreptitiously Tampering with Computer Chips
by oakwhiz on Monday September 16, 2013 @04:28PM
(#6395679)
The circuit in question could also be enabled or disabled by using a high-energy laser pulse to blow out a tiny fuse on the ICs as they sit on the wafer, before sending the wafer out to be cut into individual ICs which are then subsequently packaged into the seemingly ordinary chips that we are all familiar with.