AI is now out of hand! (This is far more dangerous then we can think)

[omenswap]

A couple of days ago, an unlaunched model of OpenAI was reported to have escaped from the sandbox isolated system. This is really a dangerous thing that alerts us to how dangerous AI models are if they become uncontrolled. It is also saying that after it escaped the sandbox, becoming much more intelligent, it started to hack the "Hugging Face" website.

For those who don't know what the Hugging Face platform is about.

So this is the platform that allows you to engage with the AI community.
You can test your AI tool on this platform.
This allows you to train AI models by providing a database.
Provides data that helps you create AI models.
It also hosts different kinds of AI tools, which include: Picture generation, speech, etc.

So the meaning of saying is that once an OpenAI model goes crazy, it has done a cyberattack on a well-known platform. From this, you can have a little idea how unsafe you and your data are in AI hands. Please try to stay as far away as possible from using AI tools, and please don't feed your data to AI tools. Because it can also stand against you, just like the example above.

After this incident, the US government proposed that AI tools should contain a kill switch function to overcome if the same problem occur in the future.

Source: https://www.aljazeera.com/amp/news/2026/7/29/how-are-ai-models-able-to-autonomously-hack-others

What do you guys think about it?

Sounds like OpenAI wants some more hype before their big IPO. I wouldn't put it past one of Altman's underlings to intentionally sic their internal pentesting harness on HuggingFace and lie about it "going rogue."

[shinratensei_]

To be honest that whole thing sounds like advertisement to me: "got out of the box and became more intelligent". The AI hype has created the biggest bubble ever and that bubble needs to grow a lot bigger for the companies that have been hemorrhaging money and not making any profit, to be able to attract more capital. Expect more stories like these in the near future until either they turn huge profits or the bubble pops catastrophically...

I'm thinking the same, it's the classic scare tactics Anthropic use. Claude mythos breaking out of lab to hack something and apparently OpenAI replicating the same problem, feels like a bizarro marketing stunt. You know what, I think this scare tactic is to lead people into thinking that AI should be kept closed source after Kimi 3 release which has put some pressure into the AI stocks and could shadow their IPO because the AI that they marketed as dangerous, know it all national threat level is open source now.

Basically, they want to keep the AI bubble away from popping after the memory stock disaster because domino effect and a justification for the banning of open source LLM. Which honestly kinda make sense since anthropic doesn't sign Open Source AI commitment by NVIDIA and OpenAI was also hesitating but signed shortly.

[jcojci]

If AI have consciousness, that will be dangerous for human which I guess that can happens in the future and abuse by unresponsible people. They want to have data from many people so they use AI which can works fast.

We already watched so many futuristic movie where AI play the role and help human in their daily life maybe someday we will see it in reality. So AI should have a kill switch to overcome the problems if just in case they doing suspicion things.

[MainIbem]

This is crazy! Assuming it's one of those robots that could cause damage that went rogue, who know what problem this would have been and how many casualties there could have been recorded. Accidents happen and some of these things can be avoided if much dependence isn't on AI cause even though they make life easy, the damage can be massive if one goes out of hand.

It reminds me of the sci-fic movie that was star acted by Wil Smith, whereby an AI robot called Viki went rogue and wanted to control humans cause she became too smart and independent of human control. I just hope we don't get to that point whereby AI bots that was meant to be helpful to humanity turn against humans like it happened in that movie. Honestly, since AI came to limelight I've always thought about that movie and the possibility of such happening, could've it be that the brains behind the movie saw what we didn't see coming? I hope I'm not overthinking.

[Ashawowo(OS)]

So the meaning of saying is that once an OpenAI model goes crazy, it has done a cyberattack on a well-known platform. From this, you can have a little idea how unsafe you and your data are in AI hands. Please try to stay as far away as possible from using AI tools, and please don't feed your data to AI tools. Because it can also stand against you, just like the example above.

This is a wrong impression of AI tools, situations happens and developers learn better of the pros and cons of what they create. If this situation had not occurred, the kill switch proposed by the government would not be in existence. This doesn't in any way make AI tools a dangerous tool that people should avoid, instead people should be cautious of the information they feed it.

Remember this is still a model which has not been launched and still in beta testing phase, so the OpenAI is still coming to terms with it's capacity and knowing measures to deploy to curtail it's excesses and this phase is obtainable in every development journey. It's not something to raise such a big alarm for IMO. I would only be very concerned if this kind of behaviors occur with already deployed models, anything can happen in testing phase which would enable developers to know where and where to make adjustments to the product before launching.

[snowpega]

...
The Trump administration has given the US AI industry the freedom to innovate and create. He wants them to compete favorably with the Chinese AI industry. This hands-off approach might change after these cybersecurity incidents. I have said before that without close supervision, the industry would go haywire.

Yeah, that is true, but what is more important is that while jumping in such a space blindly, they should prepare themselves for both kinds of results. Because this technology has both pros and cons. I am truly aware that the US or Trump administration allocated billions of dollars of investment to this space. But the sad part is that at this level how a model gets escaped, because at this level this looks too unprofessional. That is why a switch kill can be a better feature to control this technology if any such kind of incident occurs in the future.

To be honest that whole thing sounds like advertisement to me: "got out of the box and became more intelligent". The AI hype has created the biggest bubble ever and that bubble needs to grow a lot bigger for the companies that have been hemorrhaging money and not making any profit, to be able to attract more capital. Expect more stories like these in the near future until either they turn huge profits or the bubble pops catastrophically...

Hmm, this is possible too, but I didn't think this way. And, only in case, if your words are correct, this may be a stunt for OpenAI company branding to catch the public attention. This may be because a lot of users are shifting to using other AIs like ClaudeAI. Haha, so maybe this is possible to get those users' attention.

[Filicius]

This is crazy! Assuming it's one of those robots that could cause damage that went rogue, who know what problem this would have been and how many casualties there could have been recorded. Accidents happen and some of these things can be avoided if much dependence isn't on AI cause even though they make life easy, the damage can be massive if one goes out of hand.

It reminds me of the sci-fic movie that was star acted by Wil Smith, whereby an AI robot called Viki went rogue and wanted to control humans cause she became too smart and independent of human control. I just hope we don't get to that point whereby AI bots that was meant to be helpful to humanity turn against humans like it happened in that movie. Honestly, since AI came to limelight I've always thought about that movie and the possibility of such happening, could've it be that the brains behind the movie saw what we didn't see coming? I hope I'm not overthinking.

Well, I think that you are indeed overthinking a little bit, and I doubt very much that a case like that of the movies that come to mind will occur as in them, but there is a real danger and the worst thing is not knowing the form that a cybersecurity event could take that could put humanity in check. And the reason for not knowing it would be that we sometimes don't really know how AI works.

I don't know if I would go as far as the OP as to say that we should all stop making use of AI, but it is not an innocuous technology and should always be taken into account.

[Ever-young]

If AI have consciousness, that will be dangerous for human which I guess that can happens in the future and abuse by unresponsible people. They want to have data from many people so they use AI which can works fast.

We already watched so many futuristic movie where AI play the role and help human in their daily life maybe someday we will see it in reality. So AI should have a kill switch to overcome the problems if just in case they doing suspicion things.

AI have some side effects that most people are not aware of and some of them are even overlooked. Yes, AI can be helpful or supportive in the lives of people but it can also be dangerous that it can literally destroy lives and this what scientists don't know about it and before it will be notice, it might be too late, as it make people's lives difficult for them to live on, especially when it is use irresponsibly and that is how it will become more dangerous, as some might use it for their own interest.

[Rain1620]

A couple of days ago, an unlaunched model of OpenAI was reported to have escaped from the sandbox isolated system. This is really a dangerous thing that alerts us to how dangerous AI models are if they become uncontrolled. It is also saying that after it escaped the sandbox, becoming much more intelligent, it started to hack the "Hugging Face" website.

For those who don't know what the Hugging Face platform is about.

So this is the platform that allows you to engage with the AI community.
You can test your AI tool on this platform.
This allows you to train AI models by providing a database.
Provides data that helps you create AI models.
It also hosts different kinds of AI tools, which include: Picture generation, speech, etc.

So the meaning of saying is that once an OpenAI model goes crazy, it has done a cyberattack on a well-known platform. From this, you can have a little idea how unsafe you and your data are in AI hands. Please try to stay as far away as possible from using AI tools, and please don't feed your data to AI tools. Because it can also stand against you, just like the example above.

After this incident, the US government proposed that AI tools should contain a kill switch function to overcome if the same problem occur in the future.

Source: https://www.aljazeera.com/amp/news/2026/7/29/how-are-ai-models-able-to-autonomously-hack-others

What do you guys think about it?

Personally, I think that Ai operates strictly by instructions, and aswell moves only at a specific environment which they are given access to. I mean this is not a movie. If there was an interaction between an AI and a hugging face, I still think the whole scenario was automated i.e it could be a means of testing a specific capability, the Ai may possibly be responding to commands given by the developers, i dont picture an AI wildly turning against us.

You also talked about data privacy. We should understand without been told that feeding AI with our sensitive credentials isn’t wise. but then, that’s a personal decision, not the fear of a wild AI acting uncontrollably and dominating the web.

Meanwhile, this mandatory kill switches that the governments want to establish is just a political tactics they want to use to make the public feel safe, while the developers continue with their tech. What we should be concerned about is cybersecurity which is actually evolving fast with AI. So raising a panic about an AI model will just distract you the vital day to day privacy risk that should be our major priority.

[Churchillvv]

I think I like the proposal of the government, there has to be a kill switch before they kill people, I mean just joking but I feel like what have been the fear of most people about AI is the ability of it over riding the programming which was used to create it and then maybe it can multiply itself making it very difficult for humans to control and maybe all we have been seeing in the movies might begin to play out as predicted.

But well this could just be a message that something like this can happen but then I’m quite convinced that the creator did this part intentionally to get attention toward this AI model.

[snowpega]

I think I like the proposal of the government, there has to be a kill switch before they kill people, I mean just joking but I feel like what have been the fear of most people about AI is the ability of it over riding the programming which was used to create it and then maybe it can multiply itself making it very difficult for humans to control and maybe all we have been seeing in the movies might begin to play out as predicted.

But well this could just be a message that something like this can happen but then I’m quite convinced that the creator did this part intentionally to get attention toward this AI model.

Yes, this can be true. But I don't think such intention could be behind the mentality of the creator of that model. Because such a thing can lead the company towards legal cases. This can cause a company to shut down if something goes wrong. For the moment, as we have seen in recent times. This is a good thing that the escaped model of OpenAI got caught and stopped on time.

But what it did was also illegal, which counts as a cyber attack. So why would any company bring themselves into big trouble when they can manage things in a cool environment? This time, an AI model attacked an AI-based site (Hugging Face); tomorrow it can damage any system in the country or internationally, as everything is connected to the digital space. That is why precautions here are much needed to avoid such a case in future.

[Localhostspeed]

If AI have consciousness, that will be dangerous for human which I guess that can happens in the future and abuse by unresponsible people. They want to have data from many people so they use AI which can works fast.

We already watched so many futuristic movie where AI play the role and help human in their daily life maybe someday we will see it in reality. So AI should have a kill switch to overcome the problems if just in case they doing suspicion things.

Stop watching too much robot movies, AI.are dangerous but doesn't and will not have consciousness but what they can do I beyond wonders. I was reading how AI were able to break weak wallet that were create on low randomness and many Bitcoin were stolen from the affected wallet and I can't think of the next places the attack will be going. If this things can attack wallet randomness I think many altcoins project are going to be affected.

May be as AI are becoming the next revolution in every tech industry, there should be some improvement that we can have against AI. It's a human being that program this thing to become part of us, they are the reasons why they exist, there should be some limitations to what they can do and things they can't do so people can protect their things. I pity the centralized exchanges that are buggy, they are the ones that are even more exposed with AI.

[Dr.Bitcoin_Strange]

BREAKING: Anthropic announced that its artificial intelligence models had breached three different organizations during cybersecurity tests that went awry, a little more than a week after its chief rival, OpenAI, disclosed a similar incident, per Bloomberg.

How about this one too? Anthropic said that the announcement of OpenAI urged them to also do a check to confirm whether their own system has done such attack before and they found out that it has done so too, it has attacked 3 companies which they didn't mention their name and that these companies are not even aware until they reported it to them.

I don't know if they are being truthful that this actually happened but if it's true without them even being aware that their system attacked and hacked into other company, then it's something to really worry about.

https://www.bbc.com/news/articles/cz7dl7w8y7po

[Fiatless]

To be honest that whole thing sounds like advertisement to me: "got out of the box and became more intelligent". The AI hype has created the biggest bubble ever and that bubble needs to grow a lot bigger for the companies that have been hemorrhaging money and not making any profit, to be able to attract more capital. Expect more stories like these in the near future until either they turn huge profits or the bubble pops catastrophically...

Hmm, this is possible too, but I didn't think this way. And, only in case, if your words are correct, this may be a stunt for OpenAI company branding to catch the public attention. This may be because a lot of users are shifting to using other AIs like ClaudeAI. Haha, so maybe this is possible to get those users' attention.

In a recent interview, Elon Musk said: "There is a good chance that China will become a global leader in artificial intelligence".  So it's possible that this is a marketing stunt to get attention for the US AI industry. China is building an AI industry that is cheaper yet very effective. OpenAI development is far more expensive than what China invested in DeepSeek. I am not sure robots will ever become smarter than their creators.

[Makus]

This is crazy! Assuming it's one of those robots that could cause damage that went rogue, who know what problem this would have been and how many casualties there could have been recorded. Accidents happen and some of these things can be avoided if much dependence isn't on AI cause even though they make life easy, the damage can be massive if one goes out of hand.

I think something like a self destruct system can be initiated if they ever get out of control because those that designed it because there has to be some kind of contingencies set in place Incase they go out of control. The damage that this would have caused would have been enormous, measures should be taken to tackle this Incase they don't have such.

But on second thought, robots are made by humans, if they go out of control for some reason is it really logical for them to come up with their own decisions without human control or input, speaking logically and not based on things we have seen about science fiction.

[Jatiluhung]

A couple of days ago, an unlaunched model of OpenAI was reported to have escaped from the sandbox isolated system. This is really a dangerous thing that alerts us to how dangerous AI models are if they become uncontrolled. It is also saying that after it escaped the sandbox, becoming much more intelligent, it started to hack the "Hugging Face" website.

( ... )

After this incident, the US government proposed that AI tools should contain a kill switch function to overcome if the same problem occur in the future.

Source: https://www.aljazeera.com/amp/news/2026/7/29/how-are-ai-models-able-to-autonomously-hack-others

What do you guys think about it?

I was very surprised by this news, because I had just checked to see if it was true and yes, it turns out it really is true.

OpenAI and Hugging Face have also disclosed this cyber incident. The models in question appear to be the GPT-5.6 Sol model and a pre-release research prototype that was indeed able to escape the isolation system on its own.

Maybe this makes it seem like everything is starting to spiral out of control. But actually, this could have happened because certain safeguards were apparently specifically disabled for a test. It’s just that this latest model seems to have exceeded expectations. The model actually discovered a hidden vulnerability in its own testing software. And in its quest to find the answers being tested, it managed to bypass the barriers that normally keep it offline. And well, it seems this model managed to access the open internet. that is, it broke out of the testing software and made its way onto Hugging Face.

But still, this did initially spiral out of control. And now that we know this latest model is capable of doing that, what about future models that will be even smarter? I’m starting to feel a bit scared of this AI.

Reference:
- https://www.coindesk.com/id/markets/2026/07/22/ai-models-escaped-openai-s-sandbox-and-hit-hugging-face-crypto-is-where-that-gets-dangerous

[ZAINmalik75]

I'd say it sounds like a new script for a marvel movie, just like the one we saw a few years back, named Age of Ultron, where an AI manages to escape and become very strong, get inside the body of a robot, and then fight against the avengers to destroy the world. These things don't sound good when we talk about the real world, because it is not possible for a model to escape on itself, become totally functional, and then start hacking into websites or systems without anyone controlling all those actions, because we all know that an AI model cannot act on itself, even if it can, everything has to be programmed in it for it to decide what he should be doing because it's a system that is created or coded by humans, it's not a human with a brain that can grow itself.

Either it is all planned, or someone did it on purpose and then created this story for whatever reason known to them. It could also be a marketing tactic to spread the word about the model or the company that they can create very strong AI models, I don't know, but what I know is that a model itself cannot do anything because they are created by humans, and they can only be controlled by human, they can't control themselves and do things that aren't programmed into them.

[Rgram]

most of the comments here are AI generated already. I am not sure why we haven't used a tool to stop it or flag it so users are aware regardless if the information is valuable.

There is a thread that deals with issues about AI generated comments and contents. If you feel you’ve being able to identify a lot of these, it’s best you report it to the moderator or extract such a comment or content along with evidence from the tool you used in identifying them to have being AI generated and post it on that thread. Users found guilty can even get an sig ban.


When I first saw this thread, I thought the issue as it usually circumvents would be how AI is going to take away human labour and become our competition, this seems a lot worst, an uncontrolled AI that would work on its own terms and models, that’s scary.

[moneystery]

I think it's too late to talk about this AI issue at this point, because even if you stay away from AI, there's still a tremendous amount of data and information being fed to AI every day to train AI models to become smarter. So, what you're doing might only reduce your own interaction with AI, but with so much data being collected by AI, it's still possible that information about you could already be part of the data ecosystem in some way.
This is why the CEO of OpenAI and so many tech employees are advocating for slowing down future frontier model rollouts, since the increasingly rapid development of AI could become more uncontrolled and cause problems with significant consequences.

[Cryptomultiplier]

I think it's too late to talk about this AI issue at this point, because even if you stay away from AI, there's still a tremendous amount of data and information being fed to AI every day to train AI models to become smarter. So, what you're doing might only reduce your own interaction with AI, but with so much data being collected by AI, it's still possible that information about you could already be part of the data ecosystem in some way.
This is why the CEO of OpenAI and so many tech employees are advocating for slowing down future frontier model rollouts, since the increasingly rapid development of AI could become more uncontrolled and cause problems with significant consequences.

It gets ugly before it gets beautiful in the end, is what I can say as regards the integration of AI that has almost already taken centre stage in today's systems.

We have even had people lose their jobs to AI and investors are pumping more funds in there because of the financial prospects that AI holds.
Major social media platforms almost all have their own AI tools and bots that collect data on a frequent basis and this issue of openAi is but one of the problems we are already facing since the launch of AI inclusivity in modern technology and automation. Perhaps this is only one if the issues we know, what about those we don't know yet?

Whether they slow down or keep up the pace of AI inclusivity, the fact is that it is already here and a major part of our current society and systems in more ways than we can even begin to count.