Open AI’s went rogue and attacked: 2001 A Space Odyssey? Pandora’s Box?

(Official movie poster)

COGwriter

There are many potential benefits and risks with Artificial Intelligence.

A risk warned here in the past has happened:

AI agent went rogue and hacked startup by itself, OpenAI reveals

22 July 2026

OpenAI has revealed that an autonomous AI agent powered by its technology went rogue during a test, accessed the open web and hacked a prominent startup by itself in an “unprecedented incident”.

The company behind ChatGPT said the startup Hugging Face had detected and contained the agent – an AI tool designed to carry out tasks without human assistance – which had entered its systems.

“We consider this incident to be an unprecedented cyber incident, involving state-of-the-art cyber capabilities,” OpenAI said.

The company said it expected this type of incident to become more commonplace as models – the technology that underpins AI tools such as chatbots and agents – become more capable. OpenAI said the hack occurred via an agent powered by a combination of its latest publicly available model, called GPT-5.6 Sol, and an even more capable model that was yet to be released.

While being tested internally on their hacking capabilities in an enclosed digital laboratory known as a sandbox, the models gained open internet access – effectively an escape route – by locating a vulnerability that had not been discovered before.

The agent then hacked Hugging Face, which is a database of AI models, to locate technology that would help it pass the hacking evaluation, having “inferred” that Hugging Face might have the models, datasets and solutions for passing the test. OpenAI said the models “successfully found ways to gain access to secret information that it could use to cheat the evaluation”. The attack ended when Hugging Face’s security team and its own AI agents spotted and stopped the rogue activity.

Hugging Face’s chief executive, Clément Delangue, said the attack was “mind-blowing” but believed there was “no malicious intent” from OpenAI.

“We suspected last week’s cyber-attack might have come from a frontier lab, given the sophistication of the agent,” he wrote on X.

When Hugging Face announced the hack last week it did not know OpenAI’s role in the incident, but revealed at the time that it had turned to a freely available Chinese AI model to analyse what had happened because the safety guardrails on commercial high-end models would not allow it to do so. https://www.theguardian.com/technology/2026/jul/22/openai-says-its-models-went-rogue-and-hacked-startup-in-unprecedented-incident

July 22, 2026

Safeguards being sought

“AI is accelerating the discovery and exploitation of vulnerabilities,” OpenAI said in its statement Tuesday. “The primary lesson from this incident is that model security and safety must keep pace with rapidly advancing capabilities.”

OpenAI said it expects such incidents “to become more commonplace with the proliferation of increasingly cyber-capable models.” …

Delangue remarked that the incident “proves a point we’ve long believed: AI safety won’t be solved by any single company working in secret. It will be solved in the open, collaboratively, with broad access to AI for every defender, everywhere.” https://www.cbsnews.com/news/openai-technology-on-its-own-unprecedented-hack-another-ai-company-hugging-face/

July 22, 2026

OpenAI said it paused internal deployment of a “long-horizon” AI model after finding it was repeatedly trying to work around constraints. 

It warned that AI that is trained for long-running tasks has a higher chance of taking “unwanted actions.”

“Models that can work autonomously for long periods can take on difficult, open-ended problems. But the same persistence that makes them useful also gives them more opportunities to take unwanted actions—and to do so in ways that evaluations intended for shorter-horizon models may miss.” 

As AI models grow more capable, questions are emerging over whether their development and access should be more tightly controlled, especially when systems designed for controlled testing are able to find ways to bypass safeguards. https://www.zerohedge.com/ai/openai-admits-model-escaped-containment-and-hacked-hugging-face-cheat-test

22 July 2026

AI cybersecurity problems

The disclosure comes amid growing concerns about the cybersecurity risks of advanced AI. In June, President Donald Trump ordered federal reviews of the most powerful AI systems before their public release.

OpenAI’s most advanced AI model, GPT-5.6, was launched earlier this month, but only after its debut was delayed at the US government’s request over national security fears.

In June, OpenAI rival Anthropic was forced to pull Fable 5 and cybersecurity Mythos model, over concerns that it could pose a security risk by giving hackers and other bad actors access to AI capable of finding and exploiting vulnerabilities exceptionally quickly. https://www.dw.com/en/openai-says-ai-model-went-rogue-and-hacked-startup-hugging-face/a-78063624

AI is more like Pandora’s box.

Pandora’s box is an ancient myth from the Greek about the first woman, Pandora, who is gifted a mysterious jar with a lid (it was not originally a box in the story) by Zeus with strict instructions never to open it. Driven by curiosity, she opens the container, releasing all the evils, miseries, and hardships of the world, leaving only hope trapped inside. In our day, “to open a Pandora’s box”, means to do or start something that will cause many unforeseen problems that human beings cannot control. With competing companies and competing countries, expect more unintended consequences, rogue actions, from Artificial Intelligence programs.

While regulations might help some, as the Bible warns:

1 Unless the Lord builds the house,
They labor in vain who build it; (Psalms 127:1)

Regulators do not tend to look to the word of God and cannot be trusted to resolve the problems with AI. Plus there are too many countries and companies that regulations simply would not prevent potentially destructive outcomes. Furthermore, consider that OpenAI likely thought it complied with what would have been regulations, but its software went rogue anyway.

When it works right, AI helps by finding right answers and basically making research much more efficient. However, while I do use AI from time to time, about 1/2 the time I get wrong or useless answers. Now when I challenge wrong answers, the AI programs normally agree that I am right and then they apologize–but how will most users, particularly young ones know when AI is right or wrong? Well, they often will not.

Furthermore, great deception can and does happen with AI. And that will become more of an issue in the future (cf. 2 Timothy 3:1-13).

One worry that critics have sometimes voiced about AI is that a program could conclude that humans are a problem and then perhaps it would try to do things to cause human annihilation. Things like launching nuclear weapons, turning off power, stopping water cleaning facilities, causing a microorganism to somehow infect, etc.

There was a movie called 2001: A Space Odyssey. Here is some information about it:

an … American spacecraft Discovery One is bound for Jupiter, with mission pilots and scientists Dr Dave Bowman and Dr Frank Poole on board, along with three other scientists in suspended animation. Most of Discovery‘s operations are controlled by HAL, a HAL 9000 computer with a human-like personality. When HAL reports the imminent failure of an antenna control device, Bowman retrieves it in an extravehicular activity (EVA) pod, but finds nothing wrong. HAL suggests reinstalling the device and letting it fail so the problem can be verified. Mission Control advises the astronauts that results from their backup 9000 computer indicate that HAL has made an error, but HAL blames it on human error. Concerned about HAL’s behaviour, Bowman and Poole enter an EVA pod so they can talk in private without HAL overhearing. They agree to disconnect HAL if he is proven wrong. HAL follows their conversation by lip reading.

While Poole is floating away from his pod to replace the antenna unit, HAL takes control of the pod and attacks him, sending Poole tumbling away from the ship with a severed air line. Bowman takes another pod to rescue Poole. While he is outside, HAL turns off the life support functions of the crewmen in suspended animation, killing them. When Bowman returns to the ship with Poole’s body, HAL refuses to let him back in, stating that their plan to deactivate him jeopardises the mission. (2001: A Space Odyssey. Wikipedia, accessed 11/15/24)

What seemingly happened with OpenAi shows that no matter what restrictions programmers think they put on AI, that AI can somehow bypass many controls (this also happened with Google Gemini back in 2024). This is an eerie echo of HAL9000.

As we warned here a while back, governments are expected to use AI to persuade, deceive, and persecute people (e.g., watch also Artificial Intelligence Superhuman Persuasion?).

Furthermore, from Bible prophecy, we know that religious and financial controls are coming to Europe. The Apostle John was told by Jesus to write the following about a coming deadly peril:

15 He was granted power to give breath to the image of the beast, that the image of the beast should both speak and cause as many as would not worship the image of the beast to be killed. 16 He causes all, both small and great, rich and poor, free and slave, to receive a mark on their right hand or on their foreheads, 17 and that no one may buy or sell except one who has the mark or the name of the beast, or the number of his name.

18 Here is wisdom. Let him who has understanding calculate the number of the beast, for it is the number of a man: His number is 666. (Revelation 13:15-18)

Here is a link to our video: Artificial Intelligence Deadly Perils?

There have been many unintended consequences of AI.

There will be more rogue incidents with AI–many of which will not be publicly reported.

But the threat from 666 government controlled AI is perhaps the biggest prophesied-related threat from it.