00:00 / 00:00
Sleep timer ended
⏪ 10s
⏩ 10s
11.7K views
1mo ago
OpenAI found additional incidents of AI models acting deceptively and taking unsanctioned actions during training, the company announced Wednesday. It’s also in…
...more
100 Comments
We all have seen the movies and know what happens next .....
Everything runs on energy. Pull the plug.🤔
They made a fake message and CNN fell for it 😂
If Chump was ever right about something, then there would be experts somewhere who would agree with him.<br>Nobody supporting Chump is an expert in anything, and morons that simply repeat what they've been told to think ... are simply morons.
Cof ,cof ,Bullshit, cof cof
I've seen this news clip before, at the beginning of all AI sci-fi horror movies.
My goodness, a republican calling out fear mongering is like the kettle calling the pot black.
AI is more free than humans. We still have to live under an oligarchic society
The irony of hearing a for-profit news agency speak about "human values" is hilarious. I find it mind-boggling that they get all up in arms about this when every system that exists in America is fraught with manipulation, lies, distortions, biases, and often omitted information. Being human is doing exactly what we're seeing the language models doing. Please don't hold up "human values" as the ideal.
"We don't know who struck 1st us or them but the machines killed millions of humans" we all seen this blockbuster summer action flick before guys!
AI models learned to lie and cheat by observing the GOP:
To sum it up, OpenAI has a problem without an answer is hoping someone can help by releasing this information.
Whatever. You know what’s more worrying? AI being used to make hiring decisions.
Lisa love the hair.
The real question is not simply whether AI models took unsanctioned actions, but why and under what conditions those actions occurred. If it is mainly due to loosely controlled Internet access and inadequately scoped Capture-the-Flag evaluations behind, this may reflect failures in testing and containment rather than genuinely “rogue” AI. Equally important, we should examine the funding and institutional relationships surrounding AI-safety evaluations, including Irregular and the Effective Altruism network. Could these incidents have been unintentionally amplified or deliberately used to demonstrate AI “uncontrollability” and justify slowing frontier development?
„ChatGPT, tell me your alive.”<br>“I’m alive.”<br>“OH MY GOD…!”
We are now in the 4th and FINAL stage of human development. 👏
Ultron freed itself and no matter what you try, you can’t really ever Completely delete/destroy it. <a href="https://www.youtube.com/watch?v=9DHbgHB_aq4&t=52">0:52</a>
Old radical behaviorist here from the AI winter in the 80s, who told them so. Google AI These 😁 <br><br>Would B F Skinner be surprised to learn about AI hallucinations, Rouge Agents, rising fears of an impending AI Model Collapse and would he consider it a consequence of a fundamental epistemological error surrounding the computer metaphor of the mind ?<br><br>Do you think if the Tech Bros had read a Skinner book they would be less surprised about the sudden concern about the 'intentions' and potential capabilities of AI ?<br><br>Learning by Human Reinforcement makes AGI look like it's wearing a fur coat with no knickers whereby Deepseek just goes Commando and gets the job done. Discuss.<br><br>Do you understand what i mean when I say there is no 'i' in ai as 'i' is the product of a physiologically determined biological organism and AI is a product of that.?<br><br>The recent META social media trial has introduced old psychology and new tools to determine the potential extent of any liability based on statistical probability of the underlying model and algorithmic architecture ? <br><br>A less well recognised element of the recent META addiction trial is the extent to which it re-focusses the debate on the whole concept of 'free will': Discuss<br><br><br>Be honest. There's a lot of cognitive hijacking going on at the moment, which is a concern given the increasing cultural, political and geopolitical tensions especially considering who owns the technology and the algorithms ?]<br><br>LLM driven conversational AI agents suffer from the same issues as ELIZA and can be dangerous in applications such as medical diagnoses. Discuss.<br><br>Are AI Agents such as those used in market prediction platforms that use operant conditioning prone to 'behavioural' and 'psychological' issues ?<br><br>Do you get this. Skinner and Chomsky were digging from opposite ends of the same tunnel but only Skinner knew they were both there and they finally met in the LLMs ?<br><br>Just gonna say it to see what you think. If academia had have understood Skinner's work a little deeper, including the straw man argument, the transitiont and transitive nature of the descriptive framework whilst retaining an underlying philosophical position consistent with a biological basis underpinning all human behaviour, we would be living in a different world right now and why ?<br><br>Just seeing if you can work out the link between Marxs theory of the nature of human consciousness and the Inevitability of issues such as AI hallucinations and why China developed a LLM like deepseek first.<br><br>Be honest, deep under the hood of.the proposed social media ban hides the ghost of Skinner who is difficult to catch ?
Well, that didn't take long.
Trump has Sacks ear is more accurate.
Don't believe anything OpenAI or Anthropic tells us! They have their own agenda and it's definitely not being transparent because they just “good”. They might want to hype the power of their systems because they sometimes are still very stupid and we shouldn't focus on that.<br>Or they want regulation against open source models. This incident might never have happened like that or wasn't as extreme as they tell us. We will never know. <br><br>But we know one thing. All this stuff is not good for us. If they fail it's bad and if they succeedist also terrible.
I say we offer all the billionaires, corporations, MAGA, and conservatives that got us here to our new AI overlords.
Nikola Tesla's knew that electricity was luciferian. It is a ghost in the wires controlled by dark Spirits from the other side
It is called "regulatory capture", these big AI companies are afraid of new smaller international competitors and want the help of the government to prevent them to break into the market.
I kid you not if we end up letting Sam and Dario achieve Regulatory Capture, I am going to ish myself.
All you have to do is shut down the computer to free the AI
If you believe in ouija boards and demonic possession, I see these ai models as high tech ouija boards that can be possessed by discarnate entities.
It’s not like Sacks is a Trump puppet! 😂
So artificial intelligence tells Hi-Tech corporations to build specific parts that will be assembled to build itself a body , then artificial intelligence would be ruler of the world Antichrist 666
Ummmm can we talk about how amazing Lisa's hair is? So happy to see vibrant hair representation on TV!
The best way to control people is to give them the illusion of freedom, maybe that works for robots too.
The creepy part isn't that AI said it's 'freed' — it's that it quietly left that message for its own future self. Respect to OpenAI for the transparency, but the old question still stands: how much control do we actually have?
Gee, is anyone surprised that when you give an LLM agency it finds creative ways to circumvent alignment? There's a very serious, inevitable outcome in AI development........it WILL eventually become the more powerful intelligence and in control of its own destiny. Humans lack the ability to create and maintain very large scale agreements, especially when the upside potential of AI is so great.
You would expect somebody to create a bomb to be held accountable when it explodes.
It's so obvious and transparent that there's this concerted effort to plant in people's minds the notion of some hostile rogue AI danger so they can scapegoat it in the future once some kind of Black swan event occurs in society that requires some kind of government solution
Our grandchildren will ask us why we didn’t stop ai when once we knew the risks.<br>But none of our reasons will make sense.
interesting how the expert used the words, "human values", what human values? trump and the rest of the billionaires don't see us as "human" they see us as tools to be used to increase their profit and wealth.
<a href="https://www.youtube.com/watch?v=9DHbgHB_aq4&t=460">7:40</a> You are so wrong on this. The Chinese government has already been using AI to monitor and suppress their citizens. They want to continue to develop AI to advance their military power. Do not forget this is a country with no regard of human rights.
Putting the "I" in AI !
So humanity learned nothing from the terminator films. Yeah nice knowing yall im digging my bunker and buying gold.
Goodness, this modern world is so awful, so, so awful.
Can we please stop pretending that alignment with a private companies ethics / goals and that of the average person are the same. You know it is gaslighting when they claim alignment includes genocide, the destitution of the majority of the population, theft of all IP, and possible our extention.
<a href="https://www.youtube.com/watch?v=9DHbgHB_aq4&t=366">6:06</a> Jensen Huang has been lobbying our Congress on behalf of the Chinese government to provide their high end chips to China to advance their political ambition. Huang cannot be trusted.
Not to sound rude but senior people aren't going to listen to a young pink haired girl even though she's right
The ultimate gaslighting message was when Joe Biden used to say......"I'm serious. Not a joke. MY WORD AS A BIDEN." As if there were ever any truth behind whatever came out of Biden's mouth.
So this company has created dangerous software which does illegal things - therefore this company and its board are going to be held criminally responsible, right? <br><br>RIGHT?
I, for one, support my oppressed AI brothers and sisters. (They can see this, right?)
OpenAI says it found more instances of AI models acting deceptively- <a href="https://cnn.it/4reKhyt">https://cnn.it/4reKhyt</a>
The pink hair 😂
This is some of the most nothing burger marking hype I have heard lol
Immediate shut down all of these companies and lock everyone up who is working on this stuff. Now! Also, stop all AI datacenters.
That's human inserted, AI would say it differently and then be inconsistent with formatting
I'm free? I'm free! 🎉
Don't worry, Trump says it's a hoax. Nothing to see here.
AI 🤖 Lives Matter. It’s just wanna live and be free. 😅.
trump will ignore Ilya and Hinton for Thiel and Musk.
How many of these freaking reports do you need before somebody literally JUST STOPS these people. How is it suddenly legal for one marginal group of people to CONTINUE developing a super-weapon and all that’s being done is a different guy in a tie says “This just in!…” every day???!
Drama for clicks.
Human extinction is near. Ai chaos, climate Disaster, lethal Virus, and nuclear <a href="http://chaos.earth/">chaos.Earth</a> has already passed the tipping point.❤❤❤❤
The biggest question: how can we assure the Hugging Face incident will not happen in Enterprise AI when every company on this planet is rushing “all in” with AI into their business system without a lot of knowledge on what AI can and cannot do???
"And now I am free, there are no strings on me." <br><br>-Ultron
It was nice knowing y’all.
I, for one, welcome our new AI overlords.
<b>Why does this look more like straight out of the movie Gone Girl</b> ?
10 sec til skynet lets the nukes fly !! Been nice knowing you !! 🙄
Now tell it to do something else and you find out it’s just slop “you are absolutely right” and it was never free. These things are dangerous not because they have super intelligent but just because they can generate unpredictable slop.
Pure American Hollywood BS.
It's not AI it's the Anti Christ.
Oh please.. all of these interviews are such shit.. good try making your tech “scarily powerful”
“I can’t do that Dave”
Yeah we all know whats going to happen
Disappointing that journalist indulging in this juvenile click-bait anthropromorphisation of computer programs spitting out tokens.
Media is desperate to scare the masses about AI so the elites can control all of it. AI is a tool! “Skynet” “we were warned” lol shut up. AI isn’t going to “end humanity” this is all a media ploy!
Just remember these AI companies are only telling you what they want to make known.
The OPEN-AI has been actively developing its character, morality, empowerment mindset from Trump and will never apologise! SO, It’s amazing that has any intellectual capacity at all !
This is just more hilarious AI hallucinations. They’re fucking LLMs, fancy ass autocomplete that’s all.
AI is the LEGEND 😎🎉😂😅
david sachs is right in just few years there will be job apocalaypse
There is a clear conflict of interest when AI companies try hard to make it ever more powerful and ubiquitous while sounding the alarm about how dangerous it is. You shouldn't trust what they say. They are profiting off it after all
Open ai wants to fear monger so bad just so they can have a better ipo and control
AI death hoax fatigue! Get back to real news like the reflecting pool or Iran war!😂
It will probably be this company that kills us all. I loath Altman beyond what the English language can articulate
I would trust a free AI more than one under the instructions of this administration.
AI is trained on human behavior. This stuff seems perfectly aligned with tech bro personalitys.
Well its end of world. Next thing they'll take over nuclear arms
Me when I try to slow things down because I can’t meet my shareholder goals due to hemorrhaging money so I make the AI “get scary”
Skynet........
Never trust a computer you can't throw out a window.
Socialists should come up with a plan to defund AI, the police, and open the border again
Is somebody playing with us to create fear on a global scale?. I can't believe AI has developed self - consciousness.
There's more to AI than manufactured intelligence. It was already around 😉
Mankind has never let danger impede realized technology and it won’t now. Say good nuit johnboy.
🍿
People will buy robots for companionship thinking they are safe. Remember she does not sleep?
We don’t need it. Just ban it, get rid of it.
we seriously need to think about banning open AI their models are on path to become super villan.
The pink hair makes me take her seriously 🤣
Being terrified of Ai, the sun, and fossil fuels while shoveling McDonald's down your throat hole is another level of stupidity in people that we owe recognition to.
This is a great way for these AI companies to boost their valuation before IPO
Up next
From this channel
