Bitget Hack, Duelbits, ColdCard, Haruko, Liquid - Time to change?

[EarnOnVictor]

Thanks for the correction. I double-checked now, and you guys are right, as there were recent breaches just last month. I didn't do my background digging, but relied on the old info I had, plus I didn't remember receiving a breaching emails from Bitget and Duelbits. I've adjusted my post accordingly, but still maintain the so-called "AI agent" hacking debunking."

AI hacking is clickbait in my opinion, they don't do anything better than humans do until now. Maybe they are performing low level social engineering tricks for now and we had seen one of the government website had unauthorized access from chatgpt and we also had seen the GPT 6 which can performa tasks on its own by taking the control of our devices but yet they are not on the elevel of exploiting the vulnerabilities yet.

I thought just exactly as you have said. Most of these claims, if not all, targeting high profile hacking cases have nothing to do with AI hacking. However, AI hacking is also real, and such AIs are being sold regularly in dark web and telegram links. While majority of them are also scammers targeting other scammers/hackers to dupe them. I recently read that such tools sales have even risen more than 25 times in just a year. This shows it is booming, but the credibility of the work is what I don't know.

[cryptomaniac_xxx]

And this is the responsibility of these platforms and so as we as individual users, we also need to be aware of our own security. While these platforms can be accountable if ever a small hole has been abused to them, they can potentially what is lost for their users. But we as individuals, if you are at fault there's no way that you'll get a potential refund or safu or however they call those funds recovery programs they have. We are our own security and we protect ourselves depending on the practice that we do.

With the hacked on Bitget, it's more of the responsibility of exchanges, although they said that they can covered what has been lost. But still though, that is very bad security on their end because this is not the first time that it did happened to them.

That should really be covered by them with the losses they've made through the hack. But, for how long they'll be able to? What if there's another version of this hack and they were targeted again by these hackers? I don't think that they'll be able to afford to cover the expenses and refunds of their users although they should still do it.

If I recall correctly, in the early years of some South Korea exchanges, they are not mandated to have funds at their arsenals or at least covered what have been lost in a instance of hacked. But the South Korean regulators implemented a very strict rule moving forward to have funds bye their exchanges so that they can cover everything.

So this is what really happened here, Bitget have their reserves for this hack. Although again, it's a huge amount of money that they might have to keep if they only spend more money in protecting their infrastructure and somewhat stop this hack.

[ultrloa]

AI hacking is clickbait in my opinion, they don't do anything better than humans do until now. Maybe they are performing low level social engineering tricks for now and we had seen one of the government website had unauthorized access from chatgpt and we also had seen the GPT 6 which can performa tasks on its own by taking the control of our devices but yet they are not on the elevel of exploiting the vulnerabilities yet.

I thought just exactly as you have said. Most of these claims, if not all, targeting high profile hacking cases have nothing to do with AI hacking. However, AI hacking is also real, and such AIs are being sold regularly in dark web and telegram links. While majority of them are also scammers targeting other scammers/hackers to dupe them. I recently read that such tools sales have even risen more than 25 times in just a year. This shows it is booming, but the credibility of the work is what I don't know.

Ai hacking truly exist and it help bad actors to fasten up their jobs done with their target platforms.

Notable hacking incident which been assisted by Ai.

https://en.cryptonomist.ch/2026/10/03/bitget-387m-hack-ai-tracing

https://blockonomi.com/ai-powered-attack-drains-112m-in-bitcoin-from-coldcard-hardware-wallets

https://bex.co/blog/2026/04/12/ai-trading-agent-45m-exploit-defi-security-breach

Also got curious on what you have said about ai tools which has been sold on dark web. The it seems its true because many articles tackles about it and here's one of them discuss about that incidents https://www.cybersecuritydive.com/news/ai-hacking-tools-sale-underground-forums/827807

[tabas]

That should really be covered by them with the losses they've made through the hack. But, for how long they'll be able to? What if there's another version of this hack and they were targeted again by these hackers? I don't think that they'll be able to afford to cover the expenses and refunds of their users although they should still do it.

If I recall correctly, in the early years of some South Korea exchanges, they are not mandated to have funds at their arsenals or at least covered what have been lost in a instance of hacked. But the South Korean regulators implemented a very strict rule moving forward to have funds bye their exchanges so that they can cover everything.

It doesn't need a government ruling just for any exchange to cover what potential loss they can make. But that's a good step there and after that, exchanges have followed the same ruling for their own implementation to give more security and assurance for their users if something bad happens.

So this is what really happened here, Bitget have their reserves for this hack. Although again, it's a huge amount of money that they might have to keep if they only spend more money in protecting their infrastructure and somewhat stop this hack.

There can be less damage for the hack but each of these exchanges should anticipate that they're always targeted and this incident is likely to happen at any given time. So, being prepared for the reserves is mostly for this instance.

[davis196]

What changes do you suggest? Better security? Stronger passwords? Total decentralization? AI is getting better at hacking and penetrating systems day after day. Perhaps our best bet would be to shut down the AI infrastructure completely, but AI has probably found a way to exist in servers outside of the AI data centers. I hear horror stories about a potential "killswitch" solution, if AI starts acting hostile. Shutting down all the internet and the global electricity grid. That would be totally "awesome" for the global economy and for the crypto world.

[348Judah]

Based on what I have seen, considering the recent apps and scams that are ongoing with some specific targeted wallets and centralized exchanges, we should avoid using this non-custodial wallet, and this is the first security measure we must ensure is in place.

After doing the first one, secondly, we must embrace multiple security measures to prevent hackers from getting access to our wallet and making a transaction from it; this may come with the idea of using a multi-sig wallet or passphrase as an additional security measure for any wallet we are using under non-custodial storage.

[cryptomaniac_xxx]

Based on what I have seen, considering the recent apps and scams that are ongoing with some specific targeted wallets and centralized exchanges, we should avoid using this non-custodial wallet, and this is the first security measure we must ensure is in place.

After doing the first one, secondly, we must embrace multiple security measures to prevent hackers from getting access to our wallet and making a transaction from it; this may come with the idea of using a multi-sig wallet or passphrase as an additional security measure for any wallet we are using under non-custodial storage.

I was more of thinking of using 2-out-3 multisig, as we all know if the attacker was able to get a hold of all your seed phrase, they cannot just stole it from you as a single key is not enough to authorized any transfer.

So that alone removes what all attacks has been so easy, because of the single key wallet. Of course, any backup protocol has it's risk, but at least with this kind of setup, you increase your chances against theft.

[Findingnemo]

Based on what I have seen, considering the recent apps and scams that are ongoing with some specific targeted wallets and centralized exchanges, we should avoid using this non-custodial wallet, and this is the first security measure we must ensure is in place.

After doing the first one, secondly, we must embrace multiple security measures to prevent hackers from getting access to our wallet and making a transaction from it; this may come with the idea of using a multi-sig wallet or passphrase as an additional security measure for any wallet we are using under non-custodial storage.

I was more of thinking of using 2-out-3 multisig, as we all know if the attacker was able to get a hold of all your seed phrase, they cannot just stole it from you as a single key is not enough to authorized any transfer.

So that alone removes what all attacks has been so easy, because of the single key wallet. Of course, any backup protocol has it's risk, but at least with this kind of setup, you increase your chances against theft.

Multi sig wallet avoids the single poiint of failure but even it can be exploited in the same way as these hacks but not common on personal wallets since these attacks target crypto-related platforms.

If we want to be sure the wallet will not be exp;loited at any cost then it chould be created in a device that is not connected to internet and from that moment it should never connected again, then we boradcast the transaction from another device.

[yhiaali3]

Multi sig wallet avoids the single poiint of failure but even it can be exploited in the same way as these hacks but not common on personal wallets since these attacks target crypto-related platforms.

If we want to be sure the wallet will not be exp;loited at any cost then it chould be created in a device that is not connected to internet and from that moment it should never connected again, then we boradcast the transaction from another device.

This solution is excellent for companies, but it is never suitable for centralized platforms where users trade and withdraw daily and therefore hot wallets must be connected to the internet.

Even multisig wallets have been targeted by hackers through social engineering, As happened in the Bybit breach incident, Hackers exploited "blind signing" vulnerability and misled signers into executing a malicious smart contract, resulting in the withdrawal of approximately $1.46 billion without the need to directly steal private keys.

[Findingnemo]

Multi sig wallet avoids the single poiint of failure but even it can be exploited in the same way as these hacks but not common on personal wallets since these attacks target crypto-related platforms.

If we want to be sure the wallet will not be exp;loited at any cost then it chould be created in a device that is not connected to internet and from that moment it should never connected again, then we boradcast the transaction from another device.

This solution is excellent for companies, but it is never suitable for centralized platforms where users trade and withdraw daily and therefore hot wallets must be connected to the internet.

Even multisig wallets have been targeted by hackers through social engineering, As happened in the Bybit breach incident, Hackers exploited "blind signing" vulnerability and misled signers into executing a malicious smart contract, resulting in the withdrawal of approximately $1.46 billion without the need to directly steal private keys.

Yeah, mostly social engineering is the cause for the hot wallets to be drained, exchanges should be very careful about who they are giving access to those settings, which involve fund management. Only experienced people need to be hired and also better not leave it to automation when the limits exceeds over few thousand dollars to avoid these kind of scams, which might delays the withdrawals a bit but can be rectified with hiring more people to handle it.

[greysonz]

A hack, a defection from Russia and other countries, and now a leak:
The FulcrumSec group announced on their website, fulcrumsec.vg, that they had obtained databases from several websites, including Bitjet. The data is over 10 terabytes in size: payment methods, logs, source data...
The first 200 GB portion has already been published.


https://x.com/hebi555/status/2112501797801939137

[cryptomaniac_xxx]

Based on what I have seen, considering the recent apps and scams that are ongoing with some specific targeted wallets and centralized exchanges, we should avoid using this non-custodial wallet, and this is the first security measure we must ensure is in place.

After doing the first one, secondly, we must embrace multiple security measures to prevent hackers from getting access to our wallet and making a transaction from it; this may come with the idea of using a multi-sig wallet or passphrase as an additional security measure for any wallet we are using under non-custodial storage.

I was more of thinking of using 2-out-3 multisig, as we all know if the attacker was able to get a hold of all your seed phrase, they cannot just stole it from you as a single key is not enough to authorized any transfer.

So that alone removes what all attacks has been so easy, because of the single key wallet. Of course, any backup protocol has it's risk, but at least with this kind of setup, you increase your chances against theft.

Multi sig wallet avoids the single poiint of failure but even it can be exploited in the same way as these hacks but not common on personal wallets since these attacks target crypto-related platforms.

If we want to be sure the wallet will not be exp;loited at any cost then it chould be created in a device that is not connected to internet and from that moment it should never connected again, then we boradcast the transaction from another device.

I understand, but I still think that it could be one of the best method that we can used for exchanges or even for our personal used.

But the problem with some who uses this services is that they totally rely on a 3rd party setup, as this was the way hackers was able to get at Bitget and so it was valid as per those singers and there's no way for them to see that it shouldn't be authorized.

Maybe what you are referring is air-gapped and so it might revived a total discussions about it since CryptoBilis latest attack.

[Synchronice]

I thought just exactly as you have said. Most of these claims, if not all, targeting high profile hacking cases have nothing to do with AI hacking. However, AI hacking is also real, and such AIs are being sold regularly in dark web and telegram links. While majority of them are also scammers targeting other scammers/hackers to dupe them. I recently read that such tools sales have even risen more than 25 times in just a year. This shows it is booming, but the credibility of the work is what I don't know.

What kind of AI tools are sold on darkweb? I mean, I don't think that these people on darkweb have their own AI but I do think that they might be using open-source AIs, so I wonder, do they run those AIs on their own servers and connect those AI tools that they sell via their own API? But won't that be extremely risky and leak their own info? I also don't think that they are bypassing OpenAI's and Claude's restrictions. I feel like those AI tools must be all scam.

[EarnOnVictor]

I thought just exactly as you have said. Most of these claims, if not all, targeting high profile hacking cases have nothing to do with AI hacking. However, AI hacking is also real, and such AIs are being sold regularly in dark web and telegram links. While majority of them are also scammers targeting other scammers/hackers to dupe them. I recently read that such tools sales have even risen more than 25 times in just a year. This shows it is booming, but the credibility of the work is what I don't know.

What kind of AI tools are sold on darkweb? I mean, I don't think that these people on darkweb have their own AI but I do think that they might be using open-source AIs, so I wonder, do they run those AIs on their own servers and connect those AI tools that they sell via their own API? But won't that be extremely risky and leak their own info? I also don't think that they are bypassing OpenAI's and Claude's restrictions. I feel like those AI tools must be all scam.

I read well about this some months ago, and yes, they are real, but they are majorly AI-powered and even some independent tools, so the server is not structured as you thought. They claimed they are useful as hacking tools, and some are sold on darkweb and Telegram. They can use open-source AI models or modified commercial AI services to assist with malware, phishing and cyberattacks, so the process is creative and not straightforward. However, many of the advertised ones are truly scams.

You might want to read this: AI-Powered Hacking Tool

[ultrloa]

I thought just exactly as you have said. Most of these claims, if not all, targeting high profile hacking cases have nothing to do with AI hacking. However, AI hacking is also real, and such AIs are being sold regularly in dark web and telegram links. While majority of them are also scammers targeting other scammers/hackers to dupe them. I recently read that such tools sales have even risen more than 25 times in just a year. This shows it is booming, but the credibility of the work is what I don't know.

What kind of AI tools are sold on darkweb? I mean, I don't think that these people on darkweb have their own AI but I do think that they might be using open-source AIs, so I wonder, do they run those AIs on their own servers and connect those AI tools that they sell via their own API? But won't that be extremely risky and leak their own info? I also don't think that they are bypassing OpenAI's and Claude's restrictions. I feel like those AI tools must be all scam.

I read well about this some months ago, and yes, they are real, but they are majorly AI-powered and even some independent tools, so the server is not structured as you thought. They claimed they are useful as hacking tools, and some are sold on darkweb and Telegram. They can use open-source AI models or modified commercial AI services to assist with malware, phishing and cyberattacks, so the process is creative and not straightforward. However, many of the advertised ones are truly scams.

You might want to read this: AI-Powered Hacking Tool

Since there are so many articles pointing out about AI tools which has been sold in Darkweb. Then we can assume that its really true.

Also in addition with the article you shared here. I want to share another interesting article about hacking tool you've mentioned sold in that area.

https://www.cybersecuritydive.com/news/ai-hacking-tools-sale-underground-forums/827807

https://www.packetlabs.net/posts/hacking-tools-sold-dark-web

https://www.dexpose.io/malicious-ai-on-the-dark-web

With existence of this, much better for people working on those sites to upgrade their knowledge as part of their defense. Because Ai hacking and other related things will continue to exist for many more year, then it could probably affect those platform didn't improve their site security.