The author did an excellent job explaining what an evil maid attack is, but a very poor job of explaining how their proposal mitigates such attack.
I think the classic "Detecting unauthorized physical access with beans, lentils and colored rice" [0] approach is simpler to understand and simpler to implement. It doesn't rely on any hardware, such as a Raspberry Pi or otherwise technology which can be more easily subject to scrutiny via Ken Thompson's "Reflections on Trusting Trust".
That's cool. I hadn't heard of that, before. I had a related idea for achieving plausible deniability of the key in full disk encryption or similar scenarios. The password would be derived from the position of sensitive, yet innocuous, elements on the device, ensuring that the seizure of the device would likely corrupt this relationship. For instance, a series of N-sided dice could be placed in specific positions on top of the device (in the case of a desktop computer, perhaps), and the password derived from their sequence. Consideration must also be given to the possibility of the device being photographed—likely from a single angle—before being moved. So, the dice would be positioned to include some amount of occlusion. Any dice-based algorithm would need to ensure the search space for the resulting key was sufficiently large.
Thanks for the feedback. My guess is that the part about destroying the random secrets is easier to understand, but the later part about a key pair and how its signing of the photo log can help with a persistent network outage is harder to understand? It does need a specific mental picture to see how it makes sense. I'll try to have more diagrams to explain.
But yeah the "random mosaic" with rice and beans is a great defense. My view is that these together can form a defense in-depth.
This reminded me of the (real life) story of Oleg Gordievsky, the FSB officer who was a double agent for the west[1]. He was alerted to the fact that the FSB were on to him and had been in his apartment because there were three locks on his front door but he never locked one of them as he didn’t have the key. He came home one day to find all three were locked.
[1] read “The spy and the traitor” by Ben Mackintyre. It’s incredibly gripping and at times hard to believe the courage and perseverance of the people involved but it was real.
And if that tickles your fancy, "Tinker Tailor Soldier Spy" [0] is an excellent miniseries from 1979 about a mole in MI6, perhaps the best spy series ever made. I didn't care about the movie much, so don't let this deter you, but Alec Guiness as George Smiley is a perfect match. John Le Carré thought so, too.
Oh and Patrick Steward plays "Karla" the soviet mastermind in this series and its successor "Smiley's People". Just a few seconds, but very memorable, its incredible really.
I like the way you made me think! It had not thought about it until now, but I take your point.
While thinking about it, this phrase occurred to me: “silver bullets are a defense against zombies.” It is not the same phrase structure as the original, but it also has the double-negative vibe, yet it feels more reasonable to me than “…are a defense for zombies”, which to me suggests that zombies would employ them against their enemies.
I think the resolution here is that defense is inherently against something, so these phrases are not unequivocally double negatives - though I also agree with nine_k’s point about a better way to say it.
EDIT: Duh! The fact that defense is inherently against something is precisely what makes these phrases look like double negatives! The resolution must be something else - maybe agreement in mood or sentiment…
Instead of deleting the secret on trip, and requiring a re-arm, it could instead derive a new secret on trip, by e.g. hashing the previous secret. That way you don't have to manually re-arm it, and you get a record of all trips.
Say e.g. a bug walks in front of the camera, tripping it. Then 1 hour a later an evil maid comes in and tampers with the system. In my design, you could look at the photo record, see that the 1st trip was a false alarm, then continue looking at the data, and see that the 2nd trip was something real.
Compared to with the current design, the bug would trip it, then you would get no record of the actual evil maid. You would see the photos of the bug tripping it, and think "oh, it's just a false alarm, I don't need to worry", and trust the computer, even though it's tampered with.
3 problems I can think with my idea are: (1) it makes the tripping less noisy, so it increases the chance someone might ignore or miss the trip. I guess with the right UX that can be mostly sovled. (2) if a bug walks in front of the camera, is that 1 trip or multiple trips? The bug would be visible for multiple frames, so it might do a ton of secret rotations for a single incident, which could present an odd UX to the user. (3) in the original design, there's an asymmetric key that's deleted on trip, which isn't really possible in my design. That means in the original design, if the phone is hacked, that doesn't let the attacker forge security footage, because the phone only has a public key, whereas in my design, if the phone is hacked, that does let the attacker forge security footage, because security is based on a symmetric key/secret. (One thing I don't understand about the original design is why it has both a symmetric key/secret and an asymmetric key. If they're both deleted at the same time, and don't auto-rotate, I don't see what benefit the symmetric key/secret provides.)
One idea to improve the (2) problem is to instead of only rotating the secret on trip, rotate for every frame, regardless of whether a trip is ongoing or not. So if there are 10 photos/sec that would be 10 rotations/sec. And then there can be a boolean in the signed data with each frame (signed e.g. with a MAC using the secret) that indicates whether there's an ongoing trip or not (and also include a timestamp in the signed data). So that means regardless of whether it's tripping, an attacker can never backdate images prior to when the attacker got control of the system.
I’ve slowly been working on building a Honeywell burglar alarm panel (a Vista15P/20P) into part of a Pelican case for travel. I can just stick up sensors where I need them temporarily (a PIR, a glassbreak, a couple motions), and then use an ECP bus decoder (like the old AlarmDecoder board[1]) to kick notifications and alerts out where they need to go with an LTE-connected miniPC/Pi.
When I need to secure an area (eg, vending at a convention at a hotel, locking up the room with stock), I can just pop down the Pelican, plug in the keypad (which doubles as the RF transceiver), stick up sensors, and I’m off to the races.
This isnt a tripwire. This is a canary. You have to actively check a canary. A tripwire would send notifications in real time without the user needing to check.
An evolution of this would be to put a server on a different network, a remote location, and have it pump out warnings the moment movement was detected and/or contact with the "tripwire" system was lost.
But the best way of preventing evil maid attacks remains knowing your hardware. Anyone trying to swap out my laptop, or open it, is going to have a problem replicating my scratch marks, my non-standard OS boot screen, or prying out the glue holding in the ram modules (to prevent cold boot attacks).
> A tripwire would send notifications in real time without the user needing to check.
c.f.
> > If any motion is detected by RPi's camera module or motion sensor, the server will delete those secrets immediately, in addition to sending push notifications to the web client.
It sends notifications in real time and tries to stay irrevocably tripped.
I was sure I'd made a comment like this before, but I'd love some sort of home-spun setup like this: https://news.ycombinator.com/item?id=2465687 ...hood, tuck, john. (2x local, 1x remote) which constantly rotated roles as to who was primary/secondary.
Basically core "chaos-infra" for your home setup(s). Hood/Tuck switch between primary and secondary, always trying to stay in touch with "John" (offsite), maybe like a primitive etcd for home automation/monitoring/backup/file-serving. Green==3good, Yellow=degraded[local|remote], Red=single-point-of-failure, Black=off/not-serving.
Other funsie to think about is getting a thumbprint/PIN-locked USB-drive to hold/unlock `~/.passwordstore/*.gpg` so that even on power-outage/reboot you'd need to physically "re-auth" to unlock important secrets.
Something like this would fit nicely into this (imaginary) setup!
I had a professor once ask about the strip of duct tape across the back of my brand new laptop. "Well, thieves cannot pawn electronics with cracked cases. So all my laptops have at least some tape so they think it may be cracked." The next lecture, the prof had a strip of masking tape on his laptop too.
But slap a tux logo and an "i l9ve truecrypt" banner on you device and nobody short of the NSA would even attempt a maid attack.
>Well, thieves cannot pawn electronics with cracked cases
Can't, or they'll get less money? I'm also not sure if I ever saw a laptop with a cracked case before, not to mention macbooks are the most recognizable and can't have cracked cases (because they're aluminum), and other laptops aren't worth stealing because their value drops sharply.
>But slap a tux logo and an "i l9ve truecrypt" banner on you device and nobody short of the NSA would even attempt a maid attack.
truecrypt is actually very susceptible to evil maid attacks because it doesn't use secureboot/tpm, which means all a baddie has to do is installed a backdoored version of truecrypt and wait for you to enter the password.
The stickers are just a statement that the owner is privacy aware. And, physically, stickers are hard to replicate quickly, preventing simple swapping of hardware. A clean iPad that looks brand new is indistinguishable from any other ipad that the maid can swap in.
The threat is them implanting some sort of device that sniffs your password next time you type it. Then they come back at a later date for the password and unlock your gadget.
The person you replied to didn't explain the full concept - it's not just nail polish, it's nail polish with glitter in to create a unique pattern that the attacker wouldn't be able to replicate.
Unfortunately... I've seen a video of somebody defeating this concept before, not by trying to recreate the pattern with new nail polish and glitter, but by using a chemical (I can't remember what) that lets them, gently and very carefully, remove the whole layer of nail polish in one piece rather than having to break it apart, and then afterwards they stuck it back in place such that it looked identical. So it's not as secure an idea as it's often considered to be.
Edit: actually my memory was slightly wrong. The video I was remembering wasn't about defeating glitter in nail polish on a screw, but about "tamper proof" stickers which are made for the same purpose. I don't know for sure if nail polish could equally be defeated, but I suspect so. Here's that video (LockPickingLawyer defeating a tamper proof sticker): https://youtube.com/watch?v=xUJtqvYDnkg&
The bullet point stating that tripwire was built for "High-ranking officials in businesses/organizations" should be removed, because that group is very unlike the "Developers of critical software", "Investigative journalists", and "Attorneys with high-profile clients" which are also mentioned.
Everybody who had the pleasure to work with "high-ranking officials in businesses/organizations" knows that this group is the one who overrides many technically optimal decisions and thinks internal policies do not apply to them. Their lives are not affected if a device is compromised because they are financially stable and can just blame an intrusion on the IT team.
The author did an excellent job explaining what an evil maid attack is, but a very poor job of explaining how their proposal mitigates such attack.
I think the classic "Detecting unauthorized physical access with beans, lentils and colored rice" [0] approach is simpler to understand and simpler to implement. It doesn't rely on any hardware, such as a Raspberry Pi or otherwise technology which can be more easily subject to scrutiny via Ken Thompson's "Reflections on Trusting Trust".
[0] https://dys2p.com/en/2021-12-tamper-evident-protection.html
That's cool. I hadn't heard of that, before. I had a related idea for achieving plausible deniability of the key in full disk encryption or similar scenarios. The password would be derived from the position of sensitive, yet innocuous, elements on the device, ensuring that the seizure of the device would likely corrupt this relationship. For instance, a series of N-sided dice could be placed in specific positions on top of the device (in the case of a desktop computer, perhaps), and the password derived from their sequence. Consideration must also be given to the possibility of the device being photographed—likely from a single angle—before being moved. So, the dice would be positioned to include some amount of occlusion. Any dice-based algorithm would need to ensure the search space for the resulting key was sufficiently large.
Thanks for the feedback. My guess is that the part about destroying the random secrets is easier to understand, but the later part about a key pair and how its signing of the photo log can help with a persistent network outage is harder to understand? It does need a specific mental picture to see how it makes sense. I'll try to have more diagrams to explain.
But yeah the "random mosaic" with rice and beans is a great defense. My view is that these together can form a defense in-depth.
Thanks for sharing again, I saw this at some point but lost the reference, great technique, cheap, easy, fun. This is art
With beans and colored rice, a smart evil maid will just wait until they next earthquake to compromise your devices.
It's vacuum packed so movement such as that of an earthquake would have no effect.
I'm surprised vaccum packing a Laptop with lentils/rice doesn't crack its screen.
Just so you know, this name is already taken by a famous security product for intrusion detection.
https://en.wikipedia.org/wiki/Tripwire_(company)
https://en.wikipedia.org/wiki/Open_Source_Tripwire
Agreed, it's pointlessly confusing to call it tripwire.
Thanks for the note. Maybe I can rename it to Tripwire AEM where AEM stands for anti evil maid.
Yep... first big project I worked on (as a baby intern). Spaff is a legend.
as well as https://en.wikipedia.org/wiki/Tripwire ;)
This reminded me of the (real life) story of Oleg Gordievsky, the FSB officer who was a double agent for the west[1]. He was alerted to the fact that the FSB were on to him and had been in his apartment because there were three locks on his front door but he never locked one of them as he didn’t have the key. He came home one day to find all three were locked.
[1] read “The spy and the traitor” by Ben Mackintyre. It’s incredibly gripping and at times hard to believe the courage and perseverance of the people involved but it was real.
And if that tickles your fancy, "Tinker Tailor Soldier Spy" [0] is an excellent miniseries from 1979 about a mole in MI6, perhaps the best spy series ever made. I didn't care about the movie much, so don't let this deter you, but Alec Guiness as George Smiley is a perfect match. John Le Carré thought so, too.
Oh and Patrick Steward plays "Karla" the soviet mastermind in this series and its successor "Smiley's People". Just a few seconds, but very memorable, its incredible really.
[0] https://www.imdb.com/title/tt0080297/
"Smiley's People" remains one of my favorite shows.
If you track it down, I highly recommend watching it with headphones. The sound design is amazing.
The sound of an empty room being profoundly menacing.
Wholeheartedly second that. Both series are amazing.
I guess this is actually not an anti evil maid defense.
It's rather an anti evil maid tool, or an evil maid defense. :)
sorry for being pedantic, but with the arms race within cybersecurity, "anti something defense" sounds like double negation to me.
I would call it "a defense against evil maid attacks" to avoid any ambiguity.
I like the way you made me think! It had not thought about it until now, but I take your point.
While thinking about it, this phrase occurred to me: “silver bullets are a defense against zombies.” It is not the same phrase structure as the original, but it also has the double-negative vibe, yet it feels more reasonable to me than “…are a defense for zombies”, which to me suggests that zombies would employ them against their enemies.
I think the resolution here is that defense is inherently against something, so these phrases are not unequivocally double negatives - though I also agree with nine_k’s point about a better way to say it.
EDIT: Duh! The fact that defense is inherently against something is precisely what makes these phrases look like double negatives! The resolution must be something else - maybe agreement in mood or sentiment…
For a second I thought Tripwire, Inc.[0] had risen from the dead with a new IDS.
[0] https://en.wikipedia.org/wiki/Tripwire_(company)
Instead of deleting the secret on trip, and requiring a re-arm, it could instead derive a new secret on trip, by e.g. hashing the previous secret. That way you don't have to manually re-arm it, and you get a record of all trips.
Say e.g. a bug walks in front of the camera, tripping it. Then 1 hour a later an evil maid comes in and tampers with the system. In my design, you could look at the photo record, see that the 1st trip was a false alarm, then continue looking at the data, and see that the 2nd trip was something real.
Compared to with the current design, the bug would trip it, then you would get no record of the actual evil maid. You would see the photos of the bug tripping it, and think "oh, it's just a false alarm, I don't need to worry", and trust the computer, even though it's tampered with.
That is honestly a fantastic idea. Many thanks for it. And I don't see any problems to fit it into the design right now.
3 problems I can think with my idea are: (1) it makes the tripping less noisy, so it increases the chance someone might ignore or miss the trip. I guess with the right UX that can be mostly sovled. (2) if a bug walks in front of the camera, is that 1 trip or multiple trips? The bug would be visible for multiple frames, so it might do a ton of secret rotations for a single incident, which could present an odd UX to the user. (3) in the original design, there's an asymmetric key that's deleted on trip, which isn't really possible in my design. That means in the original design, if the phone is hacked, that doesn't let the attacker forge security footage, because the phone only has a public key, whereas in my design, if the phone is hacked, that does let the attacker forge security footage, because security is based on a symmetric key/secret. (One thing I don't understand about the original design is why it has both a symmetric key/secret and an asymmetric key. If they're both deleted at the same time, and don't auto-rotate, I don't see what benefit the symmetric key/secret provides.)
One idea to improve the (2) problem is to instead of only rotating the secret on trip, rotate for every frame, regardless of whether a trip is ongoing or not. So if there are 10 photos/sec that would be 10 rotations/sec. And then there can be a boolean in the signed data with each frame (signed e.g. with a MAC using the secret) that indicates whether there's an ongoing trip or not (and also include a timestamp in the signed data). So that means regardless of whether it's tripping, an attacker can never backdate images prior to when the attacker got control of the system.
> NEVER PLUG/UNPLUG THE CAMERA MODULE, THE PIR SENSOR, OR WIRES WHEN THE RPi IS POWERED ON!!!
Why?! Will it will trigger W.O.P.R. and start attempting to brute force missile silo keys?
I don't know if it will cause problems. I'm just playing it safe :)
It will trigger SW bugs.
For high sec people, they should have an internal sec camera system. They are have come down in price over time
I’ve slowly been working on building a Honeywell burglar alarm panel (a Vista15P/20P) into part of a Pelican case for travel. I can just stick up sensors where I need them temporarily (a PIR, a glassbreak, a couple motions), and then use an ECP bus decoder (like the old AlarmDecoder board[1]) to kick notifications and alerts out where they need to go with an LTE-connected miniPC/Pi.
When I need to secure an area (eg, vending at a convention at a hotel, locking up the room with stock), I can just pop down the Pelican, plug in the keypad (which doubles as the RF transceiver), stick up sensors, and I’m off to the races.
[1] http://www.alarmdecoder.com/
This isnt a tripwire. This is a canary. You have to actively check a canary. A tripwire would send notifications in real time without the user needing to check.
An evolution of this would be to put a server on a different network, a remote location, and have it pump out warnings the moment movement was detected and/or contact with the "tripwire" system was lost.
But the best way of preventing evil maid attacks remains knowing your hardware. Anyone trying to swap out my laptop, or open it, is going to have a problem replicating my scratch marks, my non-standard OS boot screen, or prying out the glue holding in the ram modules (to prevent cold boot attacks).
> A tripwire would send notifications in real time without the user needing to check.
c.f.
> > If any motion is detected by RPi's camera module or motion sensor, the server will delete those secrets immediately, in addition to sending push notifications to the web client.
It sends notifications in real time and tries to stay irrevocably tripped.
I was sure I'd made a comment like this before, but I'd love some sort of home-spun setup like this: https://news.ycombinator.com/item?id=2465687 ...hood, tuck, john. (2x local, 1x remote) which constantly rotated roles as to who was primary/secondary.
Basically core "chaos-infra" for your home setup(s). Hood/Tuck switch between primary and secondary, always trying to stay in touch with "John" (offsite), maybe like a primitive etcd for home automation/monitoring/backup/file-serving. Green==3good, Yellow=degraded[local|remote], Red=single-point-of-failure, Black=off/not-serving.
Other funsie to think about is getting a thumbprint/PIN-locked USB-drive to hold/unlock `~/.passwordstore/*.gpg` so that even on power-outage/reboot you'd need to physically "re-auth" to unlock important secrets.
Something like this would fit nicely into this (imaginary) setup!
I had a professor once ask about the strip of duct tape across the back of my brand new laptop. "Well, thieves cannot pawn electronics with cracked cases. So all my laptops have at least some tape so they think it may be cracked." The next lecture, the prof had a strip of masking tape on his laptop too.
But slap a tux logo and an "i l9ve truecrypt" banner on you device and nobody short of the NSA would even attempt a maid attack.
>Well, thieves cannot pawn electronics with cracked cases
Can't, or they'll get less money? I'm also not sure if I ever saw a laptop with a cracked case before, not to mention macbooks are the most recognizable and can't have cracked cases (because they're aluminum), and other laptops aren't worth stealing because their value drops sharply.
>But slap a tux logo and an "i l9ve truecrypt" banner on you device and nobody short of the NSA would even attempt a maid attack.
truecrypt is actually very susceptible to evil maid attacks because it doesn't use secureboot/tpm, which means all a baddie has to do is installed a backdoored version of truecrypt and wait for you to enter the password.
The stickers are just a statement that the owner is privacy aware. And, physically, stickers are hard to replicate quickly, preventing simple swapping of hardware. A clean iPad that looks brand new is indistinguishable from any other ipad that the maid can swap in.
new CPUs have built in memory encryption with random key. activate it for an additional layer on top of your glue
it's called TSME on AMD
Or "memory guard". Its only available on "Pro" CPUs though, not all of them.
Perfect addition to my Darknet traphouse!
How does an evil maid get past a locked iPhone or laptop? It’s really not that easy with a proper password and encryption right?
The threat is them implanting some sort of device that sniffs your password next time you type it. Then they come back at a later date for the password and unlock your gadget.
Assuming you are a high value target I’m sure you can open a laptop and put some kind of keylogger to track the keypresses.
For a laptop with full disk encryption, there is the possibility of the attacker tampering with the hardware.
Fun project, but the chances that this is ever used successfully are exactly zero.
We used to put nail polish on all the screws/panels so that if they were ever removed it was clear as day.
If you just have nail polish, can't an attacker just put on new nail polish after removing all the previous one?
The first comment here https://news.ycombinator.com/item?id=46244062 links to something more elaborate with nail polish.
The person you replied to didn't explain the full concept - it's not just nail polish, it's nail polish with glitter in to create a unique pattern that the attacker wouldn't be able to replicate.
Unfortunately... I've seen a video of somebody defeating this concept before, not by trying to recreate the pattern with new nail polish and glitter, but by using a chemical (I can't remember what) that lets them, gently and very carefully, remove the whole layer of nail polish in one piece rather than having to break it apart, and then afterwards they stuck it back in place such that it looked identical. So it's not as secure an idea as it's often considered to be.
Edit: actually my memory was slightly wrong. The video I was remembering wasn't about defeating glitter in nail polish on a screw, but about "tamper proof" stickers which are made for the same purpose. I don't know for sure if nail polish could equally be defeated, but I suspect so. Here's that video (LockPickingLawyer defeating a tamper proof sticker): https://youtube.com/watch?v=xUJtqvYDnkg&
The bullet point stating that tripwire was built for "High-ranking officials in businesses/organizations" should be removed, because that group is very unlike the "Developers of critical software", "Investigative journalists", and "Attorneys with high-profile clients" which are also mentioned.
Everybody who had the pleasure to work with "high-ranking officials in businesses/organizations" knows that this group is the one who overrides many technically optimal decisions and thinks internal policies do not apply to them. Their lives are not affected if a device is compromised because they are financially stable and can just blame an intrusion on the IT team.
Haha, but I can still see from your statement that even though they are unwilling to follow good practices, they should in principle.