stuffoverflow

Basically they first need to get a remote shell and are then able to replace the extension source with the modified one.

This article does a good job explaining it more in depth.[0]

0: https://securelist.com/the-bluenoroff-cryptocurrency-hunt-is...

Yandex also has a pretty extensive cache, although recently they seem to have disabled caching for reddit. Otherwise it is good for finding deleted stuff, I've seen cached pages go as far back as a couple of years for some smaller/deleted websites.

From what I've seen of the people involved with IA and the related communities, I feel confident that in the very unlikely case of IA having to shut down, the data would be backed up elsewhere in its entirety.

I guess you are talking about the web interface. I don't think it is a fair way to measure the usefulness of the models because they are clearly instructed to be overly "safe". I think ChatGPT currently has nearly 2000 tokens worth of trash polluting the system prompt where as claude.ai is way too obsessed about things like copyright.

The APIs of all of the models are more permissive and refusals to answer are much more rare.

They estimated the average tax liability of noncompliers to be between $200 and $1087. Also wouldn't the fact that Norway requires to declare wealth in crypto no matter how small the amount likely inflate the number of noncompliers quite a bit? In most countries you are only create a taxable event when exchanging to fiat or other crypto.

IMO there should be more reasonable limits to how much you are allowed to sell before being liable for taxes. The amount of time it takes to properly file crypto taxes can be multiple hours especially if you do on-chain trading and need to report every invidual shitcoin exchange and so on.

I feel like a significant amount of the tax office's resources would be freed by just allowing something like 5€k a year to be sold tax free.

Since you didn't get a serious answer yet... Yes, VPNs typically use their own DNS and NordVPN is no different. As far as I've seen, the copyright trolls in Europe always go for the lowest hanging fruit which are the IP addresses of residential users. I imagine IP addresses belonging to servers or VPNs are basically disregarded.

Why not just make another google account that you solely use for registration in services like these? Use a virtual machine with VPN if you really do not want it to be linked with your real account.

It is a bit of extra work but that's just how it is nowadays.

Huggingface has quite a few spaces with the base dev model as well as different variations of it.

https://huggingface.co/spaces?sort=trending&search=dev

After reading that python discussion I have a very hard time believing the suspension was warranted unless there are some major factors i'm missing. It looks like a relatively cordial discussion yet in the suspension notice his posts are framed as having created "an atmosphere of fear, uncertainty and doubt".

I can't believe the offensive word / slur that is mentioned is the word "slut", which was said in reference to a pypi package that was removed back in the day. He even censored the word and made the SNL reference so that people would know what the word was. So according to the suspension notice the following comment was showing lack of empathy to the community members, which is just wild to me: "It was about kicking a package off of PyPI, because its docs repeatedly used a word that Dan Akroyd used to apply to Jane Curtin (brilliant comic actors from America’s “Saturday Night Live” TV show, but back in the days when it was genuinely funny".

Ok so he basically made a hidden javascript based miner and tools to distribute it.

Couldn't find anything to support the claim that he would have tried to dox Krebs. Also "maintaining biggest ransomware crypto coin network" feels like a dishonest phrasing trying to make it sound like he had something to do with ransomware. Monero was practically never used for ransomware payments back when coinhive was active, and even today Bitcoin is the most used method for ransom payments by far. Monero was simply the most profitable coin to mine with CPU.

That being said I agree that I wouldn't trust any software made by this guy. Even the hidden miner was obviously highly unethical and probably illegal.

My understanding is that at least in recent years Cellebrite has not had any 100% reliable way to crack a phone. But because people typically use short, numbers only passwords it is enough to be able to bypass the rate limiting to brute force all possible combinations.

One day I was searching live videos of a local band from youtube and when sorting by upload date the oldest video was from 2010. I knew for a fact there had to be older videos so I got a youtube API key and searched via the API, ended up finding multiple videos starting from 2006. Learned that youtube is full of videos that are basically impossible to find with the regular search.

On windows the location of chrome's extensions is "AppData\Local\Google\Chrome\User Data\Default\Extensions". You can read the source code of all of your installed extensions there. This requires you to install the extension first. It is also possible to download the crx file of any extension from the chrome web store and just unzip it to inspect the source, though i'm not sure how to do it with the official chrome. Ungoogled chromium downloads the crx file if you press "add to chrome" and then cancel.

Clicking the URL bar once always fixes it for me.

Channel names are unique but the channel display name is not. If you click on the channel you will see that the channel name is "@TesIa-Musk" and its identifier is "UC4NV-QGBa5ZtMxAvc-xxSRw", which is an unique ID that persists channel name changes. Searching historical youtube metadata for that ID reveals that the channel used to contain some random Indian songs [0]

The youtube accounts used in these scams are typically from victims of info stealer malware that has compromised their google/youtube cookies and / or passwords. It is baffling how often these Tesla or Elon musk scams gain algorithmic traction and at worst it can take hours until they are deleted. At the very least they could add some warnings if the name of a high sub count channel was recently changed.

0: https://filmot.com/channel/UC4NV-QGBa5ZtMxAvc-xxSRw

I think the biggest problem with crypto is the difficulty to get reliable information about pretty much anything. The vast majority of crypto related "discussion" on twitter or reddit is worthless. You have to dig pretty deep in to various developer communities (node, blockchain analysis, mev bot developers) to find people who actually have at least some basic understanding of smart contract interactions and can better assess the risks of different projects.

I don't share the view that Terra or FTX are good reasons to give up on crypto. Terra's promise of 15-20% interest on staking a "stable coin" should have raised big red flags. Even at the time there were notable people warning of risks in such algorithmic stable coin, but their voices were drown out under the massive hype parade on twitter that then spread on to various news outlets and so the stage was set for regular investors to lose their money.

FTX on the other hand was harder to predict but anybody following the age old mantra "Not your keys, not your coins" was safe from that fiasco.

On a purely technical level I think it is incredible what Ethereum has succeeded to achieve. A decentralized ecosystem with decentralized exchanges, loan systems, staking, and so on. Scams, shitcoins are an unfortunate side-effect but I don't see how a decentralized system could ever avoid those.

I personally use crypto whenever possible and think it is absolutely the easiest and cheapest way to transfer money to anywhere in the world. At its best it can also be a very convenient way to make online purchases.

The title is incorrect because the claude demo video does not contain a "copyrighted youtube video". It contains a webgl graphics demo with very similar fractal graphics as the author of that tweet made a year ago [0]. In the claude demo the graphics react to mouse position, while the original does not take any input.

So my understanding is that at least one of these is likely to be true:

A. The code was used in Claude's training

B. Claude used browsing and came across the code

C. The Anthropic employee who posted that video gave Claude at least a part of that code earlier in the conversation

0: https://www.shadertoy.com/view/mtyGWy

It is just a petition - basically meant to be something for the legislators to consider. AFAIK the new "coerced consent" version of chat control has not been officially proposed yet. We'll see in the coming weeks if the EU council actually tries to push it forward, and even if they do there is a pretty high change it falls through due to being fundamentally incompatible with the constitution of many european countries and / or the parliament voting against it.

Yeah you're right, people miss use the term keylogger frequently. These kind of malware are broadly called "stealers" and usually do not involve keylogging.

Actual keyloggers tend to be rare nowadays due to them being easier to detect and the fact that in general the browser data is a more valuable target.

In my experience kagi is decent, definitely gives more forum results than google. I've found yandex to be the best at finding all kinds of forum /discussion sites.

IMO the title is a bit misleading since the bug was in the mev-boost library which isn't part of the ethereum protocol nor is it needed to validate blocks. As I understand, the "victims" were arbitrage bots that one could argue simply had their risk realized. I wonder if the brothers could have gotten away with it by just paying taxes.

Here is an in-depth post-mortem on the incident: https://collective.flashbots.net/t/post-mortem-april-3rd-202...

Earlier threads about this:

https://news.ycombinator.com/item?id=40369522

https://news.ycombinator.com/item?id=40371371

https://news.ycombinator.com/item?id=40372018

Internet archive started making their own crawls in late 1996. I believe the oldest captures in the wayback machine are from data donated by alexa starting from January 1996 [1]

1: https://archive.org/details/alexacrawls?sort=-downloads&and%...