- cypherpunks - lists.cpunks.org

Tinfoil Chat
by coderman 07 Mar '20

07 Mar '20

https://github.com/maqp/tfc Tinfoil Chat (TFC) is a [FOSS](https://www.gnu.org/philosophy/free-sw.html)+[FHD](https://www.gnu.or… [peer-to-peer](https://en.wikipedia.org/wiki/Peer-to-peer) messaging system that relies on high assurance hardware architecture to protect users from [passive collection](https://en.wikipedia.org/wiki/Upstream_collection), [MITM attacks](https://en.wikipedia.org/wiki/Man-in-the-middle_attack) and most importantly, [remote key exfiltration](https://www.youtube.com/watch?v=3euYBPlX9LM). TFC is designed for people with one of the most complex threat models: organized crime groups and nation state hackers who bypass end-to-end encryption of traditional secure messaging apps by hacking the endpoint. State-of-the-art cryptography TFC uses [XChaCha20](https://cr.yp.to/chacha/chacha-20080128.pdf)-[Poly1305](https://cr.yp.to/mac/poly1305-20050329.pdf) [end-to-end encryption](https://en.wikipedia.org/wiki/End-to-end_encryption) with [deniable authentication](https://en.wikipedia.org/wiki/Deniable_encryption#Deniable_… to protect all messages and files sent to individual recipients and groups. The symmetric keys are either [pre-shared](https://en.wikipedia.org/wiki/Pre-shared_key), or exchanged using [X448](https://eprint.iacr.org/2015/625.pdf), the base-10 [fingerprints](https://en.wikipedia.org/wiki/Public_key_fingerprint) of which are verified via an out-of-band channel. TFC provides per-message [forward secrecy](https://en.wikipedia.org/wiki/Forward_secrecy) with [BLAKE2b](https://blake2.net/blake2.pdf) based [hash ratchet](https://www.youtube.com/watch?v=9sO2qdTci-s#t=1m34s). All persistent user data is encrypted locally using XChaCha20-Poly1305, the key of which is derived from password and salt using [Argon2id](https://github.com/P-H-C/phc-winner-argon2/blob/master/argon2-spe…, the parameters of which are automatically tuned according to best practices. Key generation of TFC relies on Linux kernel's [getrandom()](https://manpages.debian.org/testing/manpages-dev/getrandom.2.e…, a syscall for its ChaCha20 based [CSPRNG](https://en.wikipedia.org/wiki/Cryptographically_secure_pseudorandom…. Anonymous by design TFC routes all communication exclusively through the [Tor](https://2019.www.torproject.org/about/overview.html.en) anonymity network. It uses the next generation ([v3](https://trac.torproject.org/projects/tor/wiki/doc/NextGenOnions)) [Tor Onion Services](https://2019.www.torproject.org/docs/onion-services) to enable P2P communication that never exits the Tor network. This makes it hard for the users to accidentally deanonymize themselves. It also means that unlike (de)centralized messengers, there's no third party server with access to user metadata such as who is talking to whom, when, and how much. The network architecture means TFC runs exclusively on the user's devices. There are no ads or tracking, and it collects no data whatsoever about the user. All data is always encrypted with keys the user controls, and the databases never leave the user's device. Using Onion Services also means no account registration is needed. During the first launch TFC generates a random TFC account (an Onion Service address) for the user, e.g. 4sci35xrhp2d45gbm3qpta7ogfedonuw2mucmc36jxemucd7fmgzj3ad. By knowing this TFC account, anyone can send the user a contact request and talk to them without ever learning their real life identity, IP-address, or geolocation. Protected geolocation makes physical attacks very difficult because the attacker doesn't know where the device is located on the planet. At the same time it makes the communication censorship resistant: Blocking TFC requires blocking Tor categorically, nation-wide. TFC also features a traffic masking mode that hides the type, quantity, and schedule of communication, even if the network facing device of the user is hacked. To provide even further metadata protection from hackers, the Internet-facing part of TFC can be run on [Tails](https://tails.boum.org/), a privacy and anonymity focused operating system that contains no personal files of the user (which makes it hard to deduce to whom the endpoint belongs to), and that provides [additional layers of protection](https://github.com/Whonix/onion-grater) for their anonymity. First messaging system with endpoint security TFC is designed to be used in hardware configuration that provides strong [endpoint security](https://en.wikipedia.org/wiki/Endpoint_security). This configuration uses three computers per endpoint: Encryption and decryption processes are separated from each other onto two isolated computers, the Source Computer, and the Destination Computer. These two devices are are dedicated for TFC. This split [TCB](https://en.wikipedia.org/wiki/Trusted_computing_base) interacts with the network via the user's daily computer, called the Networked Computer. In TFC data moves from the Source Computer to the Networked Computer, and from the Networked Computer to the Destination Computer, unidirectionally. The unidirectionality of data flow is enforced, as the data is passed from one device to another only through a free hardware design [data diode](https://en.wikipedia.org/wiki/Unidirectional_network), that is connected to the three computers using one USB-cable per device. The Source and Destination Computers are not connected to the Internet, or to any device other than the data diode. https://camo.githubusercontent.com/be91382229596f4ae22250ed86091265e80ff6d7… [TFC data diode](https://www.cs.helsinki.fi/u/oottela/wiki/readme/data_diode.jpg) Optical repeater inside the [optocouplers](https://en.wikipedia.org/wiki/Opto-isolator) of the data diode enforce direction of data transmission with the fundamental laws of physics. This protection is so strong, the certified implementations of data diodes are typically found in critical infrastructure protection and government networks where the classification level of data varies between systems. A data diode might e.g. allow access to a nuclear power plant's safety system readings, while at the same time preventing attackers from exploiting these critical systems. An alternative use case is to allow importing data from less secure systems to ones that contain classified documents that must be protected from exfiltration. In TFC the hardware data diode ensures that neither of the TCB-halves can be accessed bidirectionally. Since the protection relies on physical limitations of the hardware's capabilities, no piece of malware, not even a [zero-day exploit](https://en.wikipedia.org/wiki/Zero-day_(computing)) can bypass the security provided by the data diode. How it works With the hardware in place, all that's left for the users to do is launch the device specific TFC program on each computer. https://camo.githubusercontent.com/fc12f37e9af2ef375c0c55ddf739e8900de383e2… [System overview](https://www.cs.helsinki.fi/u/oottela/wiki/readme/overview.png) In the illustration above, Alice enters messages and commands to Transmitter Program running on her Source Computer. The Transmitter Program encrypts and signs plaintext data and relays the ciphertexts from Source Computer to her Networked Computer through the data diode. Relay Program on Alice's Networked Computer relays commands and copies of outgoing messages to her Destination Computer via the data diode. Receiver Program on Alice's Destination Computer authenticates, decrypts and processes the received message/command. Alice's Relay Program shares messages and files to Bob over a Tor Onion Service. The web client of Bob's Relay Program fetches the ciphertext from Alice's Onion Service and forwards it to his Destination Computer through his data diode. Bob's Receiver Program then authenticates, decrypts and processes the received message/file. When Bob responds, he will type his message to the Transmitter Program on his Source Computer, and after a mirrored process, Alice reads the message from the Receiver Program on her Destination Computer. Why keys and plaintexts cannot be exfiltrated The architecture described above simultaneously utilizes both [the classical and the alternative data diode models](https://en.wikipedia.org/wiki/Unidirectional_network#Applications) to enable bidirectional communication between two users, while at the same time providing hardware enforced endpoint security: - The Destination Computer uses the classical data diode model. This means it can receive data from the insecure Networked Computer, but is unable to send data back to the Networked Computer. The Receiver Program is designed to function under these constraints. However, even though the program authenticates and validates all incoming data, it is not ruled out malware couldn't still infiltrate the Destination Computer. However, in the event that would happen, the malware would be unable to exfiltrate sensitive keys or plaintexts back to the Networked Computer, as the data diode prevents all outbound traffic. - The Source Computer uses the alternative data diode model. This means it can output encrypted data to the insecure Networked Computer without having to worry about being compromised: The data diode protects the Source Computer from all attacks by physically preventing all inbound traffic. The Transmitter Program is also designed to work under the data flow constraints introduced by the data diode; To allow key exchanges, the short elliptic-curve public keys are input manually by the user. - The Networked Computer is designed under the assumption it can be compromised by a remote attacker: All sensitive data that passes through the Relay Program is encrypted and signed with no exceptions. Since the attacker is unable to exfiltrate decryption keys from the Source or Destination Computer, the ciphertexts are of no value to the attacker. https://camo.githubusercontent.com/ae719d8f7c576a880b4d16607608780e2d1eceba… [Exfiltration security](https://www.cs.helsinki.fi/u/oottela/wiki/readme/attacks.png) Qubes-isolated intermediate solution For some users the [APTs](https://en.wikipedia.org/wiki/Advanced_persistent_threat) of the modern world are not part of the threat model, and for others, the requirement of having to build the data diode by themselves is a deal breaker. Yet, for all of them, storing private keys on a networked device is still a security risk. To meet these users' needs, TFC can also be run in three dedicated [Qubes](https://www.qubes-os.org/) virtual machines. With the Qubes configuration, the isolation is provided by the [Xen hypervisor](https://xenproject.org/users/security/), and the unidirectionality of data flow between the VMs is enforced with strict firewall rules. This intermediate isolation mechanism runs on a single computer which means no hardware data diode is needed. Supported Operating Systems Source/Destination Computer - Debian 10 - PureOS 9.0 - *buntu 19.10 - LMDE 4 - Qubes 4 (Debian 10 VM) Networked Computer - Tails 4.0 - Debian 10 - PureOS 9.0 - *buntu 19.10 - LMDE 4 - Qubes 4 (Debian 10 VM) More information [Threat model](https://github.com/maqp/tfc/wiki/Threat-model) [FAQ](https://github.com/maqp/tfc/wiki/FAQ) [Security design](https://github.com/maqp/tfc/wiki/Security-design) Hardware Data Diode [Breadboard version](https://github.com/maqp/tfc/wiki/TTL-Data-Diode-(breadboard)) (Easy) [Perfboard version](https://github.com/maqp/tfc/wiki/TTL-Data-Diode-(perfboard)) (Intermediate) [PCB version](https://github.com/maqp/tfc/wiki/TTL-Data-Diode-(PCB)) (Advanced) How to use [Installation](https://github.com/maqp/tfc/wiki/Installation) [Master password setup](https://github.com/maqp/tfc/wiki/Master-Password) [Local key setup](https://github.com/maqp/tfc/wiki/Local-Key-Setup) [Onion Service setup](https://github.com/maqp/tfc/wiki/Onion-Service-Setup) [X448 key exchange](https://github.com/maqp/tfc/wiki/X448) [Pre-shared keys](https://github.com/maqp/tfc/wiki/PSK) [Commands](https://github.com/maqp/tfc/wiki/Commands) [Update log](https://github.com/maqp/tfc/wiki/Update-Log)

1 0

Re: _oronavirus: _hread
by John Young 07 Mar '20

07 Mar '20

If these words are posted the _FO glimpsed passing the _un will _____. Moreover, ______. _edactions rule over _erbosity, er, _ule. _3. At 10:26 AM 3/7/2020, Cecilia Tanaka wrote: >On Sat, Mar 7, 2020, 10:26 John Young ><<mailto:jya@pipeline.com>jya(a)pipeline.com> wrote: >Do not post anything related to _oronavirus, _OVID-19, -019nCoV, >_ehovah, _od, _evil, _ezzlebub,Â _rump, _eth Rich, _ssange, >_ikiLeaks. _ust _TFU. _hx. > > >Sorry for asking, but why specifically these subjects? > >Thanks in advance!Â Kisses and hugs!Â <3

2 1

Breakdown of the healthcare system due to Covid.
by Steven Schear 07 Mar '20

07 Mar '20

https://threadreaderapp.com/thread/1236095180459003909.html

1 0

Wonder what Thomas Jefferson would think of FDR
by Ryan Carboni 07 Mar '20

07 Mar '20

https://en.wikisource.org/wiki/A_Summary_View_of_the_Rights_of_British_Amer… "By an act (3) passed in the 5th Year of the reign of his late majesty king George the second, an American subject is forbidden to make a hat for himself of the fur which he has taken perhaps on his own soil; an instance of despotism to which no parallel can be produced in the most arbitrary ages of British history." https://en.wikipedia.org/wiki/Wickard_v._Filburn The Court decided that Filburn's wheat-growing activities reduced the amount of wheat he would buy for animal feed on the open market, which is traded nationally, is thus interstate, and is therefore within the scope of the Commerce Clause. Oddly US history textbooks emphasize taxation without representation.

1 0

America is the greatest empire to grace the face of the Earth
by Ryan Carboni 07 Mar '20

07 Mar '20

America is the greatest empire to grace the face of the Earth. If there is evidence to the contrary, it ought to be ignored. The idea that it is a laughing matter that a senile buffoon might be the Democratic nominee is somewhat baffling considering criticisms of Trump. But no one really made arguments on principle, it is just to dunk on the other side. There are benefits to being the greatest empire, when Americans say America, people know they aren't referring to the continent, when Americans are wronged, our government redresses their concerns, when we travel abroad, everyone accepts American cash. We attract the greatest thinkers, the greatest doers, because that is America. Of course there will always be cliques of people who will think others are gross and rather not deal fairly with them, but prevailing circumstances will always dictate concerns. Wish the Spanish acronym for the US wasn't EE UU but EUA, but whatever, I'd don't speak it. https://nitter.net/matthewstoller/status/1235717812699594753#m Democratic insiders know Biden has cognitive decline issues. They joke about it. They don’t care.

3 2

Leaked evidence shows that Assange arrest came after Wikileaks publisher refused to name sources.
by Zenaan Harkness 07 Mar '20

07 Mar '20

----- Forwarded message from "Phillip Adams via Change.org" <change(a)e.change.org> ----- From: "Phillip Adams via Change.org" <change(a)e.change.org> Date: Wed, 04 Mar 2020 00:30:08 +0000 Subject: Leaked evidence shows that Assange arrest came after Wikileaks publisher refused to name sources. List-Id: <spc.222165.2.sparkpostmail.com> Phillip Adams shared an update on <em>Free Julian Assange, before it's too late. Sign to STOP the USA Extradition</em> Check it out and leave a comment: Petition Update Leaked evidence shows that Assange arrest came after Wikileaks publisher refused to name sources. Hi All, We are over 330,000 signatories. Click this link, read it, then read it again and then read it one more time. Understand the gravity of what is contained in this powerful article. May the murdered DNC staffer Mr Seth Rich R.I.P. ... https://click.e.change.org/f/a/pergscKKencb-qqu1B56Ew~~/AANj1QA~/RgRgQXuQP4…

1 1

Geokeys for private key recovery
by Philipp Angele 06 Mar '20

06 Mar '20

I created this alternative for seed phrases that allow you to memorize, verbally transfer and recover a private ECDSA key for Bitcoin and Ethereum. Please try the POC and share your thoughts: https://github.com/oscar-davids/geokeytool Geokeys for Bitcoin and Ether and other killer apps like pgp, ssh <https://github.com/oscar-davids/geokeytool#introduction>Introduction Today it is hard for anyone to recover a private key from memory. Brainkey implementations usually require the user to memorize a set of words. Most implementations want the user to remember a set of 12-24 words. While it is possible to memorize those seed phrases it is also very likely to forget them and highly unlikely that a user can quickly transfer them to another user without revealing the secret to unwanted listeners. Other key recovery methods have dependencies on either a trusted party or a decentralized application. Those leave the user at risk of losing access to their keys by either the trusted party removing your access or a bug in a dApp can lock your access as well. In this paper, we propose a private key recovery system that allows people to remember/recover it without any accessories or aids. The system needs just a location and a password to recover a key. We prove that neither the key nor the encryption are weakened by this system. The simplicity of the implementation allows us to get rid of: a middleman a smart contract a centralized recovery system an online connection As we will see further below, this empowers users to: create and memorize a key with very low chances of forgetting it quickly transfer a key verbally to another person recover their key at a low cost all without revealing the entropy in clear text. <https://github.com/oscar-davids/geokeytool#abstract>Abstract This paper proposes a key recovery method that uses a known and a partially known secret that forces the recovery to use a brute force mechanism to get to the full secret of the partial secret and that validates the results with the existence of the key on a blockchain, ledger, database... . A ECDSA or other cryptographic key pair might be built using one or many known secrets and one or many partially known secrets. Once this key has interacted with a blockchain, ledger... it is possible to recover it by creating all possible keys with the known and the partial known secret(s) by guessing the missing part(s) of the partial secret(s). To identify which was the original key to recover all created keys need to be looked up on a blockchain or ledger as only the one which has interacted with it will be traceable there. <https://github.com/oscar-davids/geokeytool#working-principle>Working principle A user generating a private key picks a precise location from a map and a password. A key generator uses the password and the location's geo coordinates as salt and digests everything with a high computational cost (in our PoC BCrypt) to create a ECDSA key-pair. The precision of the geo location needs to be known down to a meter level but the key generator will add a random and unknown decimal degree up to 1 millimeter depending on the targeted security of the key. As we will see, we generate entropy by having every square centimeter on Earth in combination with an infinite number of possible passwords. To recover the key the user will have to brute force all coordinates around the location on a scale of square meters with the known password. The location is quite vague since the user never knew which centimeter he chose from the map when he generated the key but only roughly the position. He will have to try through all possible square centimeter around the location he set. This will take a bit of time to recover the key but the chance for the user who knows the location and the password is high. For state of the art technology this will be on the scale of 1 computational day or 1$. For an attacker, the chance to guess both location and password is negligible. Even for a known password and an approximate guess of the location (i.e. to know which city) does not significantly speed up the brute force process. Since every generated key is valid, both the attacker and the user have to check all created keys to have a balance on the blockchain in order to identify the one they wanted to recover. <https://github.com/oscar-davids/geokeytool#conjecture>Conjecture Emotions are the best incorporeal storage in meatspace. The emotional connection to locations make them an easy thing to remember. One can transfer the key easily to someone without revealing the information in clear text and thus without reducing the entropy. An example: "The place where we kissed the first time and the nickname of our child when it wasn't born yet is the map to my treasure." This sentence would be enough for your partner to recover the key and it would be a low security risk to even say this in a room with many people listening. Yet the entropy of the information has been sufficiently transported. Your partner would have a big advantage at least to brute force to the example quicker as someone who might guess part of the secret like the name of the unborn child, even if he additionally can guess part of the location like the region or city. Humans store informations by attaching them to emotions. This is how our brain links information and how it prioritizes the lifetime (time to live = TTL) of information. The stronger the emotion the better we remember. The probability of guessing both location and password is low: For a mm range we are estimating the earth surface with 5,101e^+20 millimeter x n possibilities, while n is defined by the amount of tries the attacker is planning. If you consider only urban areas, take 0.1% of earth's surface. A scale in Millimeters is resulting in very strong encryption as there are 100 times more possibilities than with cm or 1e^6 times more than meter. Still the encryption for a single meter is good enough for smaller amounts. The definition hereof is done by the decimal-degree of the geo coordinates in the salt. (metric is just a rounded value to make presentation easier since the coordinates are not metric. 000.00000001 = 1.1132 mm(equator) - 1.0247 mm(northpole) If the location is known and the password isn’t, there is still a financial obstacle for the attack: The cost of the attack is proportionally higher than the recovery cost. In a wordlist attack of 10.000.000 words the attack would have 10.000.000 times the cost of the recovery itself. If a user has to spend 1$ or 1 day in computational efforts for recovery from a single password, the attack would cost 10.000.000$ or take 10.000.000 GPUs for one day with 10.000.000 passwords. If the balance of the key is below the attack cost or if the attack will not yield more than the computational efforts would yield with useful work like POW mining, the attacker is disincentivized to do the attack. If the location and your password leaked, you put yourself in a bad position and your only chance is that you find the key and move the balance somewhere else, before an attacker does. <https://github.com/oscar-davids/geokeytool#hashing-function>Hashing Function To strengthen the keys against any type of guessing or brute forcing attacks we use a double hashing with a compute intensive hashing algorithm. In our PoC we use BCrypt but any hashing function that introduces a time/cost variable will work, for example SCrypt or Argon2. We chose BCrypt for this implementation because of its long successful lifetime and the ability to sequence the hashing for parallelization. To keep the keys deterministic, which is a requirement to be able to recover them, we needed to get rid of the randomness in the hashing function in the elliptic curve and replace it with something deterministic that is hard but not impossible to guess. input1 = password input2 = coordinates in decimal degrees Remark: The definition of which variable is contained in input 1 and 2 is arbitrary, but needs to be aligned on. <https://github.com/oscar-davids/geokeytool#prepare>Prepare step1 BCrypt hashes the input1 with the salt==input2 at n rounds. This results in output1 step2 BCrypt hashes the input2 with the salt==output1 at n rounds. This results in output2 step3 ECDSA with output1 and output2==salt. This results into private key Rounds in bcrypt are defined to slow a single hash digest down to about 500ms on a state of the art GPU. Recovery GPU Time Cost (Jan 2020) 2x bcrypt 2e19 rounds on Nvidia 1080ti estimated cost is 1$ per day since this is the max amount you could make with mining. On mm^2 the cost of recovery from 9m^2 is about 104 GPU days or 104$ On cm^2 the cost of recovery from 9m^2 is about 24 GPU hours or 1$ On dm^2 the cost of recovery from 9m^2 is about 14 GPU minutes or 0.01$ On m^2 the cost of recovery from 9m^2 is about 9 GPU seconds or 0.0001$ We can estimate using Shannon's Source Coding Theorem the maximum entropy in bits that such a system can provide \lceil \log _2\left(x^10\left(36\times \:y^{z+1}-1\right)\left(2\times \:y^{z+1}-1\right)\right)\rceil == E After plugging in the values 95(charset), 10(number of chars), and (7+1)(decimaldegree of location) for x(charset), y(number of chars) and z(decimaldegree of location) respectively, we obtain an ideal value (E) of 132-bits. This value provides us with a key-space which is more than the recommended 128. \lceil \log _2\left(95^10\left(36\times \:10^{8+1}-1\right)\left(2\times \:10^{8+1}-1\right)\right)\rceil == E https://www.symbolab.com/solver/equation-calculator/%5Clceil%5Clog_%7B2%7D%… <https://www.symbolab.com/solver/equation-calculator/%5Clceil%5Clog_%7B2%7D%…> 132 bits entropy can be reached with the following password length and location quantifier: a password of 9 chars, a charset of 95 and a location quantifier of mm (8+1) a password of 10 chars, a charset of 95 and a location quantifier of cm (7+1) a password of 11 chars, a charset of 95 and a location quantifier of dm (6+1) a password of 12 chars, a charset of 95 and a location quantifier of m (5+1) Lowering chars on the password will result in loss of appropriate entropy. Lowering space on quantifier will result in weakness against wordlist attacks. <https://github.com/oscar-davids/geokeytool#key-validation>Key Validation: Calculate balance from public keys. Each created public key is used to calculate if it has a balance on the blockchain. Only the one that has a balance needs to be stored the rest can be deleted. <https://github.com/oscar-davids/geokeytool#difficulty-adjustments>Difficulty adjustments Every 4 years it is needed to reset the default rounds of security to match state of the art hashing. This also means to stay backwards compatible. You will have to do both the old default rounds and any new default rounds separately. We start in 2020 with recommended bcrypt cost of 19 (524288 rounds) <https://github.com/oscar-davids/geokeytool#recommended-storage-time>Recommended storage time: Since computers become faster over time, the time you store wealth behind such a key needs to be limited. Even so Moore's law makes it predictable when another difficulty adjustment is necessary, the recommendation is to create new keys at least every 4 years (assuming a max security decrease of times 16) and transfer funds to the new keys. Specialist hardware and parallel computation for recovery: It is possible to build specialized hardware that can perform more hashes per watt/cost than GPUs can. Last FPGAs updates on the topic promise a four times better efficiency and since BCrypt is not memory intensive, it can be further optimized. Compared to CPU performance per Watt GPUs are already yielding about 35 times the efficiency and FPGAs 144 times. If the specialized hardware one day outperforms user equipment at a few magnitudes, as it happened with bitcoin Asic mining, key recovery can be computed in parallel with a cloud computing provider that rents this specialized equipment. This will allow to further harden the difficulty of a brute force with additional rounds in BCrypt but will keep the requirements on the initial key generator to a minimum to match user devices that wont contain specialized hardware. (in other words it is ok if the single key generation on a phone takes a minute if the recovery time can also happen in a minute) The draw back on it is that one day if this hardware exist, key recovery will become increasingly difficult with own hardware and will need the scaled recovery architecture to match performance of an attacker one day. Awareness of infinite lifetime of keys: Making the keys deterministic also means encryption on anything you send will not stay secure forever.One day even so the elliptic curve cryptography might still not be broken, GPUs, FPGAs or ASICs might be fast enough to brute force to your old key and decrypt your old data. SO DON'T ENCRYPT PRIVATE STUFF WITH IT THAT SHOULD NEVER BE SEEN AGAIN! <https://github.com/oscar-davids/geokeytool#alibaba-and-the-forty-thieves>Alibaba and the forty thieves: Old stories already hint the way to this encryption scheme. In Alibaba knowing the location and the password: "Open Simsim" was enough to get access to the thieves magic cave, but forgetting only parts of the secret had the severe consequence to be stuck in the hideout. The missing memory on the secret made the hero unsuccessfully brute force through all kinds of possibilities since he forgot to remember the exact password but had still a memory on parts of its entropy which was that it was "Open (a specific grain)". Similar applies with the GeoKeys. Someone trying to recover it from parts of the entropy will have still much higher chances than someone guessing all of it. The location and password was leaked cause the thieves had bad OpSec and revealed the location by opening the secret gate without realizing they were spied at. Same applies to the GeoKeys. If you reveal where you search and what you search for someone might sniff that information and try to brute force faster than you. Your advantage is that other than the thieves you do not have to be physically present to do your search. Map data is more accurate and secure than real time GPS tracking and creation and recovery can happen completely offline. <https://github.com/oscar-davids/geokeytool#first-use-case>First use case: Of course this technique might not be secure enough to store millions of $$$ forever behind one key. For smaller amounts and limited time it works out well though. The reason I came up with this is inspired by stories I heard from refugees like my grandparents, losing every tiny bit they had when they were forced to flee from their country. Back then there was no bitcoin and they tried to hide what they had on their bodies. This tool increases the chances of wealth being well hidden from aggressors and recoverable for those in the greatest need. <https://github.com/oscar-davids/geokeytool#other-types-of-implementations>Other types of implementations: Geo locations are just one way to get to a vague secret. Other implementations are possible as long as they have enough quantification possibilities. In geokeys we use the space abstraction meter to millimeter. This can be replaced with something similar that is easy to remember for a human and comes with the ability to quantify it down to smaller and smaller chunks. One could use a time-date instead of an vague location for example, and to recover brute force down to milliseconds of the exact time of the date the key was created with. <https://github.com/oscar-davids/geokeytool#credits>Credits Thanks to my family who are very supportive with my ideas. Thanks to Jochen Mader from Germany for creating the first python poc script. Thanks to Oscar Davids from China for creating the openCL based recovery tool. Thanks to Hashcat for providing state of the art open source brute force technology. Thanks to Marc Cymontkowski and Fredrik Hocke from Germany for helping with the paper. Thanks to red4sec from Spain for guidance on entropy calculations Philipp Angele 02/2020 <https://github.com/oscar-davids/geokeytool#installation>

3 2

WhatsApp security flaw: 1000s of links to groups incl. 'Closed Groups' easily found
by Razer 06 Mar '20

06 Mar '20

"A DW journalist recently discovered WhatsApp links that lead to closed groups could be found with a simple Google search." "WhatsApp links that lead to closed groups can be found with a simple Google search — a major security flaw revealed by DW last week (link). Following social media outrage, the links were removed from Google’s search results. Despite the removal, however, publicly-available internet archives are still storing the information, as security researcher Lav Kumar has found out. He gathered and organized over 60,000 unique links, which can still be found on multiple websites. Of the 1,000 randomly selected links DW tested, 427 were active chat links. Even without actively joining a group, its title, description, image and creator's phone number are available for all. However, upon entering a group, it is possible to also see the phone numbers of up to 256 participants, as well as other information, and adding these numbers to one's contacts can reveal their names in the app. "We show all numbers in groups for people's safety that way they know who will receive their messages," WhatsApp told DW in response. Real-life danger Using this information, DW gained access to a group described as "Ministry of finance civil servants" in Indonesia, revealing the phone numbers of all 14 members. Several other groups appeared to be official support groups for the campaign of Brazilian President Jair Bolsonaro. Among the 427 active links DW examined, there were groups described to be for school classes, medical trainees, political campaigns, businesses, pornography and sex workers. Some groups included members with particularly sensitive identities, such as one chat with hundreds of members clearly labeled as an LGBTQ+ group in a Latin American country with high rates of homophobic murders. In some cases, the group image looked like amateur pornography or had titles such as "ex-wives leaked videos," raising questions of consent. (screenshot)" Also listed were potential terrorist groups and groups advertised as for sharing footage of "extreme" sexual content, including rape. A small number indicated that they were for child pornography. WhatsApp told DW that the company has a zero-tolerance policy around child sexual abuse and bans users immediately if they are found sharing content that exploits or endangers children. The platform also claims to ban approximately 250,000 accounts each month suspected of sharing exploitative images of children and relies on user reports and all unencrypted information to do so. 'Useful for terrorism chats' In response to the revelations, some Twitter users pointed out that this information could be used by authorities to track down illegal content without WhatsApp offering an official "backdoor" to encrypted content. "Of course there is a possibility that they left it open to search for problem groups," Jake Moore, a cybersecurity specialist and former Head of Digital Forensics at a British police force, told DW, "they aren’t always too keen to help law enforcement, so might have found it beneficial to offer it out to law enforcement and not openly mention it." On whether law enforcement would use this proactively to identify crime, Moore said: "Most police forces aren’t that proactive, [but] rather reactive. However, I would imagine it would be useful for terrorism chats, yet I doubt they use WhatsApp." However, investigations into far-right terrorism in Germany show that organizations have used Whatsapp to introduce members to each other. Read more: German far-right extremists met on WhatsApp, planned mosque attacks" More links: https://www.dw.com/en/whatsapp-security-flaw-over-60000-groups-still-access…

1 0

Re: cypherpunks Digest, Vol 81, Issue 14
by Philipp Angele 06 Mar '20

06 Mar '20

True. "Pure" randomness will always be preferred to provide a maximum security. If you need a key that you feel could keep your data encrypted for 50 years, geokeys would not be a good choice. However if it is short term use like with bitcoin you can do a trade of between level of key security and risk of losing the key. You would not have to make sure the encryption lasts for 50 years but only as long as you want to keep this key. > Am 06.03.2020 um 14:17 schrieb cypherpunks-request(a)lists.cpunks.org: > > > Even though it would be hard to guess a secret key, this technique reduces the security provided by a secret key made of pure random bits. > true? > >> Please try the POC and share your thoughts: https://github.com/oscar-davids/geokeytool <https://github.com/oscar-davids/geokeytool> >> >> Geokeys for Bitcoin and Ether and other killer apps like pgp, ssh

1 0

The Verge: A major new Intel processor flaw could defeat encryption and DRM protections
by jim bell 06 Mar '20

06 Mar '20

The Verge: A major new Intel processor flaw could defeat encryption and DRM protections. https://www.theverge.com/2020/3/6/21167782/intel-processor-flaw-root-of-tru…

3 3