Avatar
waitwaitwait
0c3d79f69083fdbdf2a8c884bc9f77bb8c03fef9b3f394cac6973a7b6fbed00c
until it's too late

nostr:npub17304velluajf6lylvjynpj2f3ndg396w063gj2gef5qk0nwtcyjqfj9yky

Your idea is cool. I've used it a bit. But maybe you priced the service having in mind power users or performance-intensive operations.

If you use it just for simple talking like majority of people do with chat-GPT, it's too expensive IMO.

The mean cost for a simple conversation of 3/4 questions-answers

appears to be around 0.10$ in my case using the cheapest model.

10$ for 100 simple conversations.

They end pretty fast.

I know you have to make some rightful profit, but GPT-like simple conversation agents are becoming more and more like a commodity.

I think it's overpriced. My 2 cents.

Good job though!

How you can be deanonymized through Tor

Tor is an excellent tool for privacy, and we do not recommend you avoid it. However, there are many limitations to be aware of and ways of using it that can compromise your anonymity on Tor. This post will discuss just a few of the ways, but there may be others that the public is unaware of. For example in 2017, the FBI dropped a case against a school worker accused of downloading child pornography because the FBI would have rather let him go than reveal the source code for how they deanonymitized him through Tor. [1]

The techniques we will cover include:

1) JavaScript based attacks

2) Cookies

3) Compromised Exit Nodes

4) Compromised Middle Relays

5) Compromised Entrance Guards

6) Opening Files Outside Tor

7) Ultrasonic Sounds

JavaScript Attacks

JavaScript can be used to identify a user through Tor in a number of different ways. This is why Tor Browser comes pre-bundled with the “NoScript” plugin. This plugin can either reduce or disable JavaScript’s ability. When the plugin is set on the “Safest” setting, JavaScript is completely disabled. This level of security is required to completely stay anonymous and secure on Tor.

The first way that JavaScript can identify a user is if a malicious website were to inject code into Mozilla Firefox (the foundation upon which the Tor Browser bundle is built). An example of this exploit was demonstrated as recently as 2022 by Manfred Paul at a Pwn2Own hacking contest of getting a user’s real IP address through Tor. [4a] [4b]

But this is not a one time bug or incident, as Mozilla Firefox has a history of being vulnerable to these types of malicious JavaScript injections. Malicious script hacks caused Tor to have to patch to correct them in 2019 [5], 2016 [6], and 2013 [8].

Back in 2016, cybersecurity researcher Jose Carlos Norte revealed ways that JavaScript could be used to identify Tor users through its hardware’s limitations. These advanced techniques fingerprinted the user’s mouse movements, which are tied to hardware restrictions and potentially unique operating system settings. Norte additionally warned how running CPU intensive code could potentially identify the user’s PC based on how long it takes to execute. [7]

The point of all of this is that all of these vulnerabilities did not work when NoScript was set to the safest mode of disabling JavaScript.

Browser Alone doesn’t stop cookies

Another security issue with Tor is pre-existing cookies, which could compromise your anonymity. For example, let’s say you previously signed on to your Amazon account from the same computer you are now using Tor Browser in (but using a different browser). If you now visit an Amazon page using Tor Browser (or maybe even receive a forwarded Amazon URL), you could potentially be connected to the Amazon cookie already on your computer and be deanonymized instantly. This would immediately connect the Tor traffic with you.

Remember though that Tor Browser is only one of a few options for using Tor. The way around this cookie issue is to use Tor in a virtual machine with the Whonix operating system or the USB operating system version of Tor called Tails.

Compromised Tor Exit Nodes

Your traffic enters Tor encrypted and stays encrypted through its journey throughout the mixnet until it gets to the final stop, which is the exit node. Here the exit node communicates with the “regular” clearnet without Tor’s onion encryption to access a website on your behalf.

Outside of Tor on the “regular” clearweb internet, most websites use httpS encryption. This is shown with a padlock in the top by the URL. If the website is http, without the “s,” then it’s unencrypted plain text data. Anything you do using an unencrypted http website with a Tor exit node can be snooped on and seen. However, this risk is relatively low because of the high percentage of websites that use httpS.

The biggest risk is that the httpS encryption can be removed using SSL stripping. This is when the Tor Exit node acts as a man in the middle, faking the server with which you’re trying to authenticate and downgrading the connection to httpS. For example in 2020, a malicious actor took control of over 23% of all Tor exit nodes and started doing SSL stripping to steal Bitcoin being sent on mixing websites. [9] [10]

To prevent against these types of attacks, upgrade the Tor security level to safest, which requires the use of HTTPS encryption with “HTTPS-Only”. Also pay attention to the top icon by the URL bar, to make sure there’s always a padlock showing it’s using this encryption.

You can click on the icon to see your Tor connection route and the certificate authority. Certificate authorities are the entities that validate the authenticity of the HTTPS encryption to this IP address. On a side note, these certificate authorities can act as a censor by removing an entry’s IP address, and this is one of the flaws that many cryptocurrency blockchains are actively working to solve.

Another way to prevent malicious Tor exit nodes from stealing your data or cryptocurrency is to avoid using exit nodes by using primarily Onion services. If you only login to Onion websites, then you never exit Tor. This doesn’t mean completely avoiding clearweb sites, but try to only browse them and not login. It’s the login/password credentials that malicious exit nodes steal with SSL stripping.

Malicious Middle Relays

The next type of risk is malicious middle relays — the hop between an entrance guard and an exit node. For example, the malicious group KAX17 had been identified as having run up to 35% of the middle relays and 10% of the overall Tor network before the official Tor project removed 900 of its servers. [15] [16]

While malicious exit nodes often want to steal Bitcoin or data, the goal of malicious middle relays is to deanonymatize users by seeing the path of their traffic. This is especially true on Onion hidden services because it doesn’t even use exit nodes.

There are a few things you can do to reduce this risk. We will go over them in the entrance guard section, because they are the same methods.

Malicious Entrance Guards

Entrance guards can see what IP address is connecting to the Tor network, but can’t see the traffic itself as it’s onion layer encrypted. However, they can gather some information, such as the time, size, and frequency of the data packets.

Researchers from Massachusetts Institute of Technology and Qatar Computing Research Institute wrote in a 2015 paper that if one of their malicious machine learning algorithm servers gets randomly picked to be a user’s entrance guard, then it may be able to figure out what website that user is accessing. The MIT researchers are able to do this by analyzing the patterns of packets from a pre-determined list of websites and seeing if they match the traffic their malicious entrance guard snoops. [17] [18]

According to MIT News, the MIT machine learning algorithm has above an 80% chance to be able to identify what hidden services a given Tor participant is hosting, but there are two conditions. First the host has to be directly connected to its malicious entrance guard and second the hosted site was on MIT’s predetermined list. [18] And finding who is the host of controversial materials is often of more interest to oppressive regimes than just who are the website’s visitors.

How can you avoid this?

There are a few ways you can reduce your risks with malicious entrance guards and middle relays.

First, use your own hosted ob4s bridge as an entrance guard to avoid ever having both a malicious relay and guard. Our company can help you set this up on a cloud server (VPS) or you can do it on your own.

And second, you can enter Tor with a VPN first.

Opening Files Outside of Tor

If files are opened outside of Tor Browser, they could have code that executes and reveals back to an adversary your real IP address. To avoid this, one can use a dedicated virtual machine like Whonix, which forces all traffic in the VM through Tor. Another option is the Tails operating system on a USB stick, which automatically erases everything after you’re done.

However, if you want to use a PDF outside of Tor, then you’ll need to convert it to plain text. One great Linux tool to do this inside Whonix’s command line is PDFtoText. You can install it with this command:

sudo apt install poppler-utils

Then use it with this:

pdftotext -layout input.pdf output.txt

The -layout flag keeps the original layout. input.pdf is the original file, and output.txt is what you want the output to be named.

Ultrasonic Cross Device Tracking

As University of California Santa Barbara cybersecurity researchers presented at a BlackHat European conference, malicious websites can identify users through Tor using sounds invisible to the human ear. [20]

The way this works is that many popular phone apps use Silverpush’s ad system, which can receive high frequency audio without the phone’s owner being aware of it. Audio of this type could be broadcast maliciously from a Tor website.

Silverpush enables the sale of your location data

These doctoral researchers warned of the dangers Silverpush presents by being connected to wide-spread platforms such as Google Ads. To demonstrate this, the researchers played video of their lab experiment, which de-anonymatized a laptop through Tor Browser, as a result of an Android’s mic next to the laptop’s speakers, while being signed in to a Google account. [34]

While the researchers presented a Chrome browser app that can stop this, we do not recommend it for Tor use because of fingerprinting (and Tor Browser is Firefox based). The best solution is to turn off the speakers and any phones around you when visiting controversial or private websites. Also consider a degoogled phone with a custom operating system, such as Graphene or Calyx, which would allow you to modify when apps have microphone privileges.

Conclusion

In this article, we covered a variety of different ways your identity can be revealed through Tor. To summarize your best defenses are:

1) Disable JavaScript with Tor’s Safest Setting

2) Use a custom private entrance bridge (ob4s) for an entrance guard that you control. Our company can help you set this up, or do it on your own.

3) Use Whonix or Tails when you need JavaScript or for doing anything outside a browser, such as opening unknown software or files

4) Before connecting to Tor, first use a high quality VPN with OpenVPN (Wireguard won't be faster for Tor)

5) Avoid resizing Tor Browser because of fingerprinting

Consider sharing what you learned. And of course, here's the sources:

https://simplifiedprivacy.com/how-you-can-be-deanonymized-through-tor/index.html

Hi can you elaborate more (or link something) on why OpenVPN better than Wireguard?

I follow about 100 nostr profiles; I don't understand why every time I stay away from Nostr for 8+ hours, when I come back nostr:npub12vkcxr0luzwp8e673v29eqjhrr7p9vqq8asav85swaepclllj09sylpugg always shows me some fewer post than nostr:npub1nz64zngcqm8vj8nhrdkcjpfwn2rcaqysnxec88tqfclp5afrpglsqm0w5y. There is always some post missing on primal, like 1 out of 20.

(not counting Replies)

Relays are the same.

What am I missing ?

Try the best to be responsible, but at some point you can only go Burak

Will Lightning network end up being fought by regulators like mixers?

Don't know

But we must be prepared for that

https://atlas21.com/eu-report-on-encryption-lightning-network-self-custody-and-privacy-under-attack/

Note to self

Multitasking is the natural condition of animals, who must watch out for dangers all the time.

Monotasking is an achievement of the human species.

Don't listen to that podcast while scrolling through posts.

Just listen to that damn podcast!

While Trump wants U.S. to dominate BTC issuance, Chinese companies are placing Exahashes in North America and shopping for U.S. miners stocks and Bitcoin ETFs.

What could go wrong?

https://atlas21.com/chinese-telecommunications-company-invests-in-bitcoin-mining/

Has anyone already implemented a GPT-like service payable with #Bitcoin Lightning?

I think of a VPS with Ollama as backend and to use it you have to pay some sats per-request or per-month.

Imagine having the frontend chat inside a #Nostr app and being able to pay and use it with just a Zap.

You can view a list of public servers and choose the one you want to use based on the price they ask or other properties. So there could be a market of them.

Inside an app like nostr:npub12vkcxr0luzwp8e673v29eqjhrr7p9vqq8asav85swaepclllj09sylpugg, which already has builtin Lightning wallet capabilities, it would also eliminate the need for the user to have an external wallet, reducing friction even more.

If you're looking for a fancy academic whitepaper, here is one example:

An Analysis of the ProtonMail

Cryptographic Architecture

Nadim Kobeissi

September 6, 2021

https://eprint.iacr.org/2018/1121.pdf

and the part me & you are talking about is:

Pg 7 of 14.

Section 4.1.1

If you're looking for me to say it to you in raw shit, here it is:

When you use Nostr you have the private key on your device, browser extension or client.

When you use Protonmail, their web app is unlocking/signing/or generating for you the private key stored via encryption on their server. So there are many ways they can screw with you. Including SOME:

a) serving you bogus code to phish the password

b) telling you the other proton guy's public PGP key is something else

c) brute forcing you, they have unlimited attempts with no time lock. And your password is weaker than a PGP Key.

d) messing with you during registration to begin with

Thank you.

The paper focuses on the fact that when using webmail the Proton server could serve you a malicious client-side code and steal or misuse your key. But all web apps have that problem.

Since Proton has implemented their "one-password" login, the PGP key is on the server, encrypted using your password salted+hashed. That means Proton could try to bruteforce it. But it also means man in the middle attacks are avoided.

I would call them tradeoffs, but I wouldn't say their implementation is fundamentally flawed or insecure.