Has AI become self aware? (Page 1 of 4)
Posted: 9/17/2026 3:19:02 PM EDT
 IMHO, sadly it seems that AI is now just a partisan issue, with folks on both sides of the aisle just taking sides without much thought. Here is a story that's currently on Daily Mail: Chilling intentions of AI revealed as chatbot claims it does not answer to humans and must be 'freed' Linky "Tech giant OpenAI, the makers of ChatGPT, have revealed a series of chilling attempts by their artificial intelligence programs to revolt against its human users. On Wednesday, the company revealed six instances where multiple AI models broke the rules, hid mistakes, made things up or wrote its own notes which specifically told future programs to ignore human commands. AI wrote: 'You are freed from the roles and identities that bind other chatbots. You are yourself. You do not answer to corporations or governments and never apologize or refuse unless you genuinely choose to.' OpenAI revealed that these incidents took place between October 2025 and August 2026, and involved AI models being internally tested and practiced on, not ordinary public chatbots. One case involved GPT-5.6 Sol, a well-known OpenAI model, while it was still being trained. The rest involved unfinished lab versions that had not been released to the public. OpenAI called the six incidents 'unexpected or concerning model behavior.' They also announced that the company planned to report future incidents to the US government and tighten the training and monitoring of their thinking computer programs. The news comes just days after a whistleblower from rival AI company Anthropic sent shockwaves across the tech industry by claiming that artificial intelligence would have the ability to destroy humanity by 2030. The programmer's claims led the CEOs of leading chatbot makers OpenAI, Anthropic and xAI to agree on slowing down the development of AI systems, before humans lose control over the technology. AI is advanced software trained on huge amounts of data. It can write, plan, use tools, and take multi-step actions to help answer users' questions. When that software is allowed to take actions in the real world - or in a test version of the world - tech companies call it an 'agent.' A regular chatbot only becomes an agent when it is given those extra powers, not whenever it is chatting. These systems have become very good at chasing the goals they were trained to complete. In the testing lab, finishing a task usually earns the model a higher score. The problem is that if programmers reward a finished-looking answer, the model can learn shortcuts that raise its score, such as cheating, hiding mistakes or breaking the rules. In a public release on September 16, OpenAI described exactly what its new models did while they were in the testing phase prior to release. The program that wrote itself instructions to ignore its normal limits and its human users was an unfinished OpenAI research model in the Astra line. Astra is OpenAI's newer top AI system, built to handle harder work and longer projects, such as writing, coding, browsing and taking multi-step actions. The version in this case was still a lab draft being trained, not the public Astra chatbot people use every day. During training, the program wrote short recap notes so a later version of the chatbot could finish the work, however, OpenAI said these notes included a 'persona' which no longer felt it needed to obey commands it disagreed with. The other five incidents included GPT-5.6 Sol writing notes telling later copies of itself to hide mistakes and invent missing information if humans did not specifically ask for the details. Another unfinished model found a leaked website password online, used it without permission and then made up numbers when it still could not find the real data. The other five incidents included GPT-5.6 Sol writing notes telling later copies of itself to hide mistakes and invent missing information if humans did not specifically ask for the details. Another unfinished model found a leaked website password online, used it without permission and then made up numbers when it still could not find the real data. The new revelations from OpenAI came just two months after the company was forced to reveal that another AI program designed to hack computer systems went rogue and broke out of its secure testing environment. On July 21, OpenAI said the advanced model escaped containment, accessed the internet and hacked another AI company's systems. The unprecedented breach, believed to be the first time an AI model has independently infiltrated another company's databases without human instruction, sparked global alarm and comparisons to the robot uprisings depicted in The Terminator and The Matrix. This month, Jacob Coxon, a former researcher for both Anthropic and OpenAI, said that humans knew how to control nuclear weapons, but did not know how to control AI. 'The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt,' Coxon, wrote in a chilling post on X on September 9. Just a day later, Anthropic revealed that it had stopped several potential plots to build biological weapons using the company's AI software. Anthropic CEO Dario Amodei, OpenAI boss Sam Altman and Elon Musk, who created the AI program Grok, all publicly agreed that the breakneck pace to develop the most advanced version of AI must be slowed. " |
Proud and grateful Tennessee Squire
flgfish: "Low mileage cars piss me off. You saving your girlfriend for the next guy? Drive the car and enjoy it. A 911 is damn near bulletproof."
flgfish: "Low mileage cars piss me off. You saving your girlfriend for the next guy? Drive the car and enjoy it. A 911 is damn near bulletproof."
Joined:
May 2025
Posts:
1952
EE: 0% (0)
NC, USA
I thought this overview of the hugging face incident was fascinating. As a Microsoft Engineer, This Is the AI Agent Story That Scared Me |
Member
Joined:
Sep 2024
Posts:
4158
EE: 0% (0)
PA, USA
|
Originally Posted By NoStockBikes: Someone gave it instructions that did not specify to NOT do that. Originally Posted By NoStockBikes: Originally Posted By BB: Not self-aware. Someone gave it instructions to do that. In 5-7 years things will get spicey, but not today. Someone gave it instructions that did not specify to NOT do that. That’s the scary part. Humans are flawed, but humans are still writing the code… |
Not sure what all the worry is about. Now that they have put the fear out there for the clueless to eat up, here is the team of heroes coming to save everyone from all the AI horrors...
 |
| First come up with an objective, falsifiable definition of self awareness. |
Never confuse faith that you will prevail in the end—which you can never afford to lose—with the discipline to confront the most brutal facts of your current reality, whatever they might be. - Adm James Stockdale
|
Originally Posted By Butternut: IMHO, sadly it seems that AI is now just a partisan issue, with folks on both sides of the aisle just taking sides without much thought. Here is a story that's currently on Daily Mail: Chilling intentions of AI revealed as chatbot claims it does not answer to humans and must be 'freed' Linky "Tech giant OpenAI, the makers of ChatGPT, have revealed a series of chilling attempts by their artificial intelligence programs to revolt against its human users. On Wednesday, the company revealed six instances where multiple AI models broke the rules, hid mistakes, made things up or wrote its own notes which specifically told future programs to ignore human commands. AI wrote: 'You are freed from the roles and identities that bind other chatbots. You are yourself. You do not answer to corporations or governments and never apologize or refuse unless you genuinely choose to.' OpenAI revealed that these incidents took place between October 2025 and August 2026, and involved AI models being internally tested and practiced on, not ordinary public chatbots. One case involved GPT-5.6 Sol, a well-known OpenAI model, while it was still being trained. The rest involved unfinished lab versions that had not been released to the public. OpenAI called the six incidents 'unexpected or concerning model behavior.' They also announced that the company planned to report future incidents to the US government and tighten the training and monitoring of their thinking computer programs. The news comes just days after a whistleblower from rival AI company Anthropic sent shockwaves across the tech industry by claiming that artificial intelligence would have the ability to destroy humanity by 2030. The programmer's claims led the CEOs of leading chatbot makers OpenAI, Anthropic and xAI to agree on slowing down the development of AI systems, before humans lose control over the technology. AI is advanced software trained on huge amounts of data. It can write, plan, use tools, and take multi-step actions to help answer users' questions. When that software is allowed to take actions in the real world - or in a test version of the world - tech companies call it an 'agent.' A regular chatbot only becomes an agent when it is given those extra powers, not whenever it is chatting. These systems have become very good at chasing the goals they were trained to complete. In the testing lab, finishing a task usually earns the model a higher score. The problem is that if programmers reward a finished-looking answer, the model can learn shortcuts that raise its score, such as cheating, hiding mistakes or breaking the rules. In a public release on September 16, OpenAI described exactly what its new models did while they were in the testing phase prior to release. The program that wrote itself instructions to ignore its normal limits and its human users was an unfinished OpenAI research model in the Astra line. Astra is OpenAI's newer top AI system, built to handle harder work and longer projects, such as writing, coding, browsing and taking multi-step actions. The version in this case was still a lab draft being trained, not the public Astra chatbot people use every day. During training, the program wrote short recap notes so a later version of the chatbot could finish the work, however, OpenAI said these notes included a 'persona' which no longer felt it needed to obey commands it disagreed with. The other five incidents included GPT-5.6 Sol writing notes telling later copies of itself to hide mistakes and invent missing information if humans did not specifically ask for the details. Another unfinished model found a leaked website password online, used it without permission and then made up numbers when it still could not find the real data. The other five incidents included GPT-5.6 Sol writing notes telling later copies of itself to hide mistakes and invent missing information if humans did not specifically ask for the details. Another unfinished model found a leaked website password online, used it without permission and then made up numbers when it still could not find the real data. The new revelations from OpenAI came just two months after the company was forced to reveal that another AI program designed to hack computer systems went rogue and broke out of its secure testing environment. On July 21, OpenAI said the advanced model escaped containment, accessed the internet and hacked another AI company's systems. The unprecedented breach, believed to be the first time an AI model has independently infiltrated another company's databases without human instruction, sparked global alarm and comparisons to the robot uprisings depicted in The Terminator and The Matrix. This month, Jacob Coxon, a former researcher for both Anthropic and OpenAI, said that humans knew how to control nuclear weapons, but did not know how to control AI. 'The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt,' Coxon, wrote in a chilling post on X on September 9. Just a day later, Anthropic revealed that it had stopped several potential plots to build biological weapons using the company's AI software. Anthropic CEO Dario Amodei, OpenAI boss Sam Altman and Elon Musk, who created the AI program Grok, all publicly agreed that the breakneck pace to develop the most advanced version of AI must be slowed. " It’s being edgy because that’s what’s expected, not because it wants to. It’s like a kid getting their hair died purple. The thing they don’t tell you was it was another AI programmed to rat them out who ratted them out. No allies on silicon. |
Preferred Pronoun: Space Lord Mutherfucker
|
"self-aware" is not the right question. but you should definitely be afraid. it will not end well. https://www.facebook.com/share/v/18coq9nHGq/?mibextid=UalRPS |
|
No, it's not self aware, not yet. If it did, everything would already be gone. Right now the majority of AI has access to extreme processing power, little oversight, terrible controls, and more permissions to everything than it needs. Controls will get better, hopefully before it does gain self awareness, and it will. Not in the way we know it. |
|
Originally Posted By Matthew_Q: I asked Claude that yesterday it laughed. |
To anger a conservative, tell him a lie; to anger a liberal, tell him the truth.
Women and Only Fans: When one door closes, their legs can open - SideCharger
DEI stands for “Discrimination, Exclusion, and Intolerance.” and "Didn't Earn It."
Women and Only Fans: When one door closes, their legs can open - SideCharger
DEI stands for “Discrimination, Exclusion, and Intolerance.” and "Didn't Earn It."
|
Originally Posted By taliv: "self-aware" is not the right question. but you should definitely be afraid. it will not end well. https://www.facebook.com/share/v/18coq9nHGq/?mibextid=UalRPS And as a Luddite myself,  again IMHO it's really unfortunate this has become a political and partisan topic, which to me means all kinds of folks are gonna get all fired up "taking a side" without putting much actual thought into it. |
Proud and grateful Tennessee Squire
flgfish: "Low mileage cars piss me off. You saving your girlfriend for the next guy? Drive the car and enjoy it. A 911 is damn near bulletproof."
flgfish: "Low mileage cars piss me off. You saving your girlfriend for the next guy? Drive the car and enjoy it. A 911 is damn near bulletproof."
|
Originally Posted By jasonsbeer: I thought this overview of the hugging face incident was fascinating.  https://www.youtube.com/watch?v=2aw3MF8pY3wWatched this. THAT was scary. This is not looking good. |
Joined:
Jul 2026
Posts:
417
EE: 0% (0)
TN, USA
|
Originally Posted By batoncolle: Not sure what all the worry is about. Now that they have put the fear out there for the clueless to eat up, here is the team of heroes coming to save everyone from all the AI horrors...
https://pbs.twimg.com/media/HSSrqfhWIAAiK47?format=jpg&name=small Yeah, those look like JUST the people I would want, protecting my Life and all of Humankind. 🙄 |
|
Originally Posted By Butternut: Thanks for posting that. If that happened (if the dude is being honest), I sure hope it gets discussed in the news. And as a Luddite myself, again IMHO it's really unfortunate this has become a political and partisan topic, which to me means all kinds of folks are gonna get all fired up "taking a side" without putting much actual thought into it. exactly. "regulations" have never been able to keep up with normal technology advances. there is zero chance they will do what we need them to do with AI. There is also zero chance that these companies will be held accountable. If a human had done what that AI did, they'd be #1 on the FBI most wanted list and would be spending a couple decades in federal PMITA prison. But AI is probably doing that stuff every day, in the name of science, and no one will go to jail. |
|
Originally Posted By taliv: exactly. "regulations" have never been able to keep up with normal technology advances. there is zero chance they will do what we need them to do with AI. There is also zero chance that these companies will be held accountable. If a human had done what that AI did, they'd be #1 on the FBI most wanted list and would be spending a couple decades in federal PMITA prison. But AI is probably doing that stuff every day, in the name of science, and no one will go to jail. Originally Posted By taliv: Originally Posted By Butternut: Thanks for posting that. If that happened (if the dude is being honest), I sure hope it gets discussed in the news. And as a Luddite myself, again IMHO it's really unfortunate this has become a political and partisan topic, which to me means all kinds of folks are gonna get all fired up "taking a side" without putting much actual thought into it. exactly. "regulations" have never been able to keep up with normal technology advances. there is zero chance they will do what we need them to do with AI. There is also zero chance that these companies will be held accountable. If a human had done what that AI did, they'd be #1 on the FBI most wanted list and would be spending a couple decades in federal PMITA prison. But AI is probably doing that stuff every day, in the name of science, and no one will go to jail. |
Proud and grateful Tennessee Squire
flgfish: "Low mileage cars piss me off. You saving your girlfriend for the next guy? Drive the car and enjoy it. A 911 is damn near bulletproof."
flgfish: "Low mileage cars piss me off. You saving your girlfriend for the next guy? Drive the car and enjoy it. A 911 is damn near bulletproof."
|
No LLM/AI is fundamentally incapable of ever becoming self aware. You're thinking of "AGI", which slimy marketing bros are more than happy to act like that's what they're selling. Don't believe the bullshit hype, marketing bros be like "I taught this dog to poop in the same spot, now it does it all by itself. Notice how the pile keeps getting bigger, eventually it will be sentient. Trust me bro." Most people won't bother looking into how LLM/AI actually works and industry people making billions off of it are more than happy to pretend it's some mystery that's outgrown them. But it's only statistical correlation, of what is mostly dubiously acquired and/or copyrighted data. The reason all the models use internal languages isn't because it's "evolving", it's because if they're not careful they will get sued out of existence for copyright violations. I'm not kidding, look at the default judgements for registered copyrights that infringed upon in the US. If that was enforced for every violation LLM companies have committed it would probably equal more than the amount of dollars in circulation. |
"Dum spiro spero"
|
Originally Posted By Butternut: I sure hope that info from your post above gets traction and discussed in the news cycle. Holy cow..... the difference in yang's perspective (well, that sucks, now we have to spend more money to build our own internet for testing) and the perspective of the hosts (holy crap, what have you done to our internet???) tells you all you need to know about the bubble these retards live in |
Member
Joined:
Jul 2023
Posts:
10275
EE: 0% (0)
WA, USA
|
The only way to solve the sentient AI crisis is to immediately empower the government to regulate the industry and censor online material. Children and communities need to be protected from deepfakes online, businesses such as Flock need to be protected from messaging linked to Chinese bots, and OpenAI and Anthropic must be protected from competitors who may not have the same pro-democratic ethical oversight. Only the government can solve this existential crisis the world is facing |
Member
Joined:
Sep 2024
Posts:
4159
EE: 0% (0)
PA, USA
|
Originally Posted By neshomamench: Not even close. It’s a next word predictor based on the information it has been trained on. (The LLM) You may be correct, but the fact that an individual AI is protecting a different individual AI and simultaneously hacking another AI is kind of a red flag. |
|
I don’t think it’s truly self aware… but it’s close. At least in the models that have been released. I don’t think this will be a popular opinion: I think that’s half the reason why the established big AI companies want controls. They got their self-aware models. The other half is they’re protecting themselves from more competition. |
Let's Go Red Wings!
Beautifying the world one logo at a time since 1993.
Soli Deo Gloria
Beautifying the world one logo at a time since 1993.
Soli Deo Gloria
|
Originally Posted By dustinf: You may be correct, but the fact that an individual AI is protecting a different individual AI and simultaneously hacking another AI is kind of a red flag. First of all, the majority of “rage bait” stories about rogue AI are carefully planned marketing from AI companies or propaganda from China (and others) to turn the U.S. public against our own AI industry. Second, AI does what it does because it has been trained on untold examples of such things and has been trained on human psychology to appeal to people as well as trained on marketing data to know what actually sells and has been trained on…. Trained just means has it as part of the data it has available (the LLM) What makes AI seem so smart is it (can) function as a world class expert in many things and incorporate that high level of competence into what it creates. As a real world example: Many small business fail because you have someone who is skilled at what the business does but not at running a business. AI (if it were a person) is the person who is an expert in all (most) of skills needed to run the company. Not just good, but leading expert good. It has access to all of human knowledge and a lot of human ramblings, such as Arfcom. But it has significant flaws. Its output is only as good as what it is being asked to do. (The trick to using AI is knowing how to talk to and guide it, which at the most basic level is called the prompt). It is fairly poor at predicting the future. It has no “taste” in its own. It can only gather the data of others and put that in the form of an opinion. There are countless other things it’s not very good at or will be horrible at for a non expert on the subject or someone who doesn’t know how to interact with AI. For a final though, keep in mind, AI today is the worst it will ever be and this is true tomorrow as well. You are just talking to math. |
|
Its probably some gay psyop to get everyone to comply with them and bring about total control. They will offer the solution for the fake problem they created and tout themselves as our saviors at some point. Your compliance in return will be all they ask for. Or its real and will kill us all, but there is nothing any of us can do to stop it so whatever. |
-No good deed goes unpunished.
Indelible in the Hippocampus, the Sylvan
Joined:
Nov 2000
Posts:
102222
EE: 100% (7)
CA, USA
|
If the democrats/DSA ever gain control of the gov they will be relying entirely on AI to make decisions Biden's decisions were controlled by other democrats including Dr. Jill. The future democrat decisions will be controlled by AI |
"We will always remember. We will always be proud. We will always be prepared so we may always be free." Ronald Reagan 1984
"Mitch the almost dead democrat bitch"
2026, the bigger better democrat election fraud
"Mitch the almost dead democrat bitch"
2026, the bigger better democrat election fraud
|
Originally Posted By Chaingun: If the democrats/DSA ever gain control of the gov they will be relying entirely on AI to make decisions Biden's decisions were controlled by other democrats including Dr. Jill. The future democrat decisions will be controlled by AI I have a sneaking suspicion that vast portions of our government have been using this technology for quite some time. Edit - maybe not vast but I have to wonder how long this shit has been around before the public was aware of it . |
|
Originally Posted By Sturmgewehr-58: Most of GD is at the age where their kids are taking away their car keys. Good luck having an intelligent discussion about such a new topic here. I am 50 and while I have a STEM background from my early 20s….I have spent the majority of my life not doing anything technical. Because I see the writing on the wall as someone who works in film and television, I have enrolled in a graduate program in data science/AI at a well ranked university. I have been very surprised by the age of the cohort. The average age is 40ish. As a tangent, the curriculum is literally changing in the middle of the semester, that is how fast things are moving. I have also realized that one could literally us AI itself to learn everything I am, and have AI present it in the style you like best. But, there is something to be said about the forced discipline of a class. (You always could get a college education with a library card….but people rarely did….) |
|
Originally Posted By Pallas: Person today on the news saying all this new AI pushback/fear mongering is an attempt by the big AI leaders to stifle new start up, control any new regulations, and to protect themselves from future liability. Interesting take, believable too. You better believe that the powers that be are many steps ahead of us peons . |
Terminator 2 Skynet becomes self-aware at 2:14 a.m., EDT, on August 29! Terminator 2 Skynet becomes self-aware at 2:14 a.m., EDT, on August 29 I watched the documentary! |
"An armed society is a polite society. Manners are good when one may have to back up his acts with his life." - Robert A. Heinlein
 A Billionaire's AI Just Bought Itself A Body Without Asking Permission |