Just saw a report on hacking these type units from Amazon and Google via laser modulation to mimic sound. Is it possible for the HomePod?
According to this article the original demo was with an iPad. The article doesn’t mention Homepods but it would seem they are vulnerable. At least Homekit wont let me unlock doors using Siri (or a burglar could shout through the window) so, since the laser simulates voice, I assume that is not a vulnerability.
The story on the national news showed the hackers opening a garage door, but all I saw were Amazon & Google hardware. I see there is an article on ZDNet about the hack so I’ll read it for mention of Apple’s HomePod.
This article refers to Siri but not Homepod. Note that a particular component of the microphone needs to be targeted by the laser beam. The Homepod has multiple microphones so I am not sure if it would be vulnerable. For example, it would make sense that it ignores sound that is only received via one of the microphones (but IT companies are not doing well at making sense at the moment!)
I’m tremendously impressed by the research into this hack, but it really doesn’t seem like a real-world worry. It’s a lot of gear and work to set up, not to mention likely being exposed while doing it, and it assumes that there’s some sort of sensitive command that could be given. Sure, some people have smart locks and garage door openers, but not many, so it would work only against people whose infrastructure was known. Breaking a window would be a lot easier.
And I suspect that the manufacturers are already discussing how to block it.
When I first heard of it, I thought it was the reverse of the surveillance trick using a laser to “listen” via window vibrations caused by audio waves; the laser would cause the window to vibrate creating audio waves for the unit’s microphones to copy. But it seems the laser beam has to actually hit a microphone on the unit. Couldn’t that be blocked just by a sheet of paper or cloth cover?
Good question. The strange part of this is that a microphone can be affected by light. The reporting is very brief on this – anyone seen a detailed explanation? These devices use micro-electro-mechanical (MEMS) microphones. I don’t know how light affects these but these are made of semiconductors, and unshielded silicon is photoelectric.
An aluminized coating or some aluminized mylar cover might prevent this and be baffled to allow sound to come through. Or a tin foil hat.
UPDATE: I looked at the original paper and they talk about what they did and listed various ways a signal might be injected into these systems but I did not see any speculation as to the underlying physics, nor any prior art about laser affects on a MEMS mike. The news reports I saw suggested that lasers moved the mike’s diaphragm but I don’t think that’s known. It could be photoelectric effect on the silicon generating and electrical signal.
Anyway, I am sticking with the foil hat on my Echo.