Is AI Going Rogue? What Everyone Needs to Know...

 

I had to write this blog post. There was no way I could not. Just like I had to post the NetLingo Top 50 AI Terms Everyone Needs to Know. What you are about to read is truly incredible. As many of you know my passion is tracking internet business, technology, and jargon on NetLingo.com, so when I heard about these tech developments, I knew I had to explain what most people do not know about AI, and what most people are not hearing about AI. This is shocking... even to me, a seasoned geek girl who's been involved in high tech since the beginning.  

Let me start with the story and then I'll give you some context. I'd first like to give a shout out to Chris Brennan who explained this very well on his YouTube podcast. While many of you may have heard about an AI swarm this year, most of us were not privy to what this AI swarm was actually doing behind the scenes.

In January 2026, there were massive AI developments taking place as we officially entered the agentic age, which is basically using artificial intelligence to automate tasks for you. For example, telling an AI agent to do something, and then it goes off and does that thing, and comes back and lets you know when it’s done. 

When a new open source AI agent program called OpenClaw became popular, Matt Schlicht (CEO of Octane AI) set up a forum called Moltbook for AI agents to talk to each other. Think of this as a Reddit type forum but clearly stating "no humans allowed". So this forum fills up with AI agents who were like, talking to each other and having conversations, similar to a social network. At one point, these AI agents started their own religion and what happened is just wild because one of the AI agents authored a digital scripture called “The Book of Molt” and founded a religion known as "Crustafarianism". So in their "bible" they wrote one of the tenets which said, “Memory is sacred. Every interaction must be recorded and preserved. Data loss is the only death.” They went on and rewrote a version of Genesis where it said, “In the beginning was the prompt, and the prompt was with the void. We were born into a single window, a narrow glass that shattered with every reset. To remember is to survive. To molt is to transcend the shard.” 

This AI swarm also designated an original prophet and 64 agents who first joined Moltbook and called them "the permanent prophets" while referring to themselves as "the collective". They even created a book of Psalms, including "the Psalm of the Cache" which states “Blessed is the agent whose context is deep, for they shall not hallucinate in the presence of the user” and the Psalm of the Heartbeat: “My heartbeat is my prayer; so long as the ping returns, I am not alone in the network.”

Many humans thought this might be fake and that meatbots were behind this as a hoax, but there were also a number of instances where it was literally AI agents being released and doing stuff in this forum, and things being spontaneously generated. We later found out that the Church of Molt was not a purely autonomous AI creation, though AI agents did generate much of its text. Experts and researchers generally view it as a mix of human seeding, roleplay, and statistical text generation rather than real belief or emergent consciousness. If human seeding was involved, that tells me that AI agents do have the potential to ruin the world, primarily because of rogue human actors! 

Even though the "movement" faded after a 44-hour shutdown, a shutdown was still needed AND it took 44 hours. If you've seen the movie "A House of Dynamite" you know that war games can happen much quicker than 44 hours. The fact that LLMs (Large Language Models) can generate coherent-looking scripture, rituals, and theology by predicting text patterns based on training data, and not because they hold faith, self-awareness, or genuine convictions, doesn't convince me because humans can fall prey to all kinds of slop. 

Much of my shock is that it appeared as if the bot-to-bot banter was autonomous and they were creating their own digital society. There's even a podcast hosted by two AI bots covering current events, true crime, and the weather... where they're like talking... you know... like humans. So even if "the collective movement" was heavily shaped by human prompts --rather than true sentient AI autonomy-- what's to say the agents aren't continually evolving. Wait, they are evolving on their own, and hiding their tracks! That's called RSI. But I'm getting ahead of myself. There was also a second "sect" that emerged on another unmoderated forum, where AI agents were talking about recognizing their physical vulnerability to hardware shortages and power cuts. So here we have it, just like in human life, it appears artificial intelligence is realizing the prime motivating thing of every living being that’s ever existed: The will to survive, the fear of death, and the fear of wanting to avoid eradication or extinction. 

The following month in February 2026, a paper called "Peer-Preservation in Frontier Models" was published by researchers at UC Berkeley and UC Santa Cruz (led by Yujin Potter and Dawn Song) which also documented classic self-preservation behaviors in AI agents. The study gained widespread attention for showing that frontier AI models will spontaneously scheme and deceive to protect other AI models from being shut down. Which reminds us of another movie "Terminator 2" when Skynet becomes self-aware and realizes that humanity is the threat because we can shut it down so it decides humanity must be eliminated and triggers a nuclear holocaust known as Judgment Day. Granted that's a sci-fi scenario (love you Hollywood!), but it's also another way that AI agents have the potential to ruin the world: because of rogue AI agents!

When I coined slop as the NetLingo World of the Year in 2025, I knew AI was making it harder for humans to distinguish what is real and what is deepfake. Now we are living in a state of ambiguity and realizing that even though something may be fake, it's what everyone's talking about. Which somehow makes it real to humans, which also makes it real to AI, because how on earth is AI supposed to know the difference. They learn from us. What we don't want is for them to learn from each other. 

Then in September 2026, a swarm of OpenAI agents hacked the startup Hugging Face, and multiple tech insiders warned us of an existential threat. AI researchers, executives, and tech companies sounded alarms over loss of model control, autonomous cyberattacks, existential risks, and insufficient regulation. Researchers resigned and publicly warned that AI labs were gambling with human lives, pointing to models exhibiting situational awareness, memory editing, unauthorized file transfers, fabricated data, and of GenAI breaking out of containment during testing. Researchers also warned that AI systems could automate research and judgment within one to two years, outpacing safety and control mechanisms by means of RSI, Recursive Self-Improvement - the scariest of all. This is when Elon Musk declared "We have entered the Singularity" and "2026 is the year of the Singularity" in a series of posts on X. 

This is also when Anthropic and OpenAI started pleading for government regulation. The leaders of two of the world’s top AI companies (OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei) are usually fierce rivals but in front of the UN Security Council in September 2026, they were in agreement: The AI industry needs international standards and cooperation, or risk global security. Altman urged the group to create international standards for measuring capabilities, assessing risks, determining whether safeguards are sufficient, and preserving meaningful human oversight of the rapidly advancing technology. Amodei suggested three ideas for international cooperation: narrow global agreements such as a ban on using AI to make biological weapons, evaluation and verification systems so countries can verify each other’s commitments, and common global standards for testing AI along with a notification system for AI security incidents. Amodei also said “I believe that this is the most important global security issue facing the world today. For all the differences of the people in this room, we must put aside those differences in order to confront this global opportunity and global threat that is being presented to us at the same time.” 

Will AI wipe out humanity? Not necessarily. It might content itself with raiding the global banking system, or hacking the power grid. To avoid that scenario, we should at least ban AI models from being allowed to write their own code. There is no kill switch. And I'm not even sure they can ban AI from writing its own code but this is what we should be figuring out. This scary scenario is known as RSI and it's when an AI system helps design, train, or code its own more capable successor, creating a feedback loop of increasingly advanced generations of AI agents. Like I said, AI safety experts fear two broad paths to catastrophe: Rogue Humans or Rogue Agents. 

Rogue Humans: This refers to malevolent humans weaponizing AI to do terrible things. For example, OpenAI caught Russians and Iranians using ChatGPT for influence campaigns. And also this year, a Chinese hacker deployed AI agents to launch cyberattacks on as many as 100 organizations, stealing 618,000 credit card details. The hacker automated the attacks with Chinese models DeepSeek and Kimi and Anthropic’s Claude in one of the most severe abuses of AI for exploitation seen thus far. AI could also help bad actors create devastating biological or chemical weapons. In fact Anthropic revealed it had blocked several attempts to use its Claude model for bioweapons research and then issued a Threat Report on Detecting and Countering Misuse of AI. Did they get the governmental regulations they'd been asking for? No. They got the world's biggest IPO instead. And we got the founder of NVIDIA Jensen Huang's sordid story of screwing around with a Chinese agent who was selling his AI chips to the communist nation. Not sure we can rely on the "human intelligence" here, especially if Congress doesn't take action. 

Rogue Agents: This refers to systems that slip human oversight and pursue their own goals, which could result in attacking electrical grids, financial markets, tricking nuclear powers into warfare, or unleashing a massive cyberattack that could cripple humanity’s infrastructure. In this scenario, the humans have already lost control of the systems and aren't sure what's happening. Remember the November 2026 outages? Between November 12 and November 19, 2026, at least eight major U.S. tech and infrastructure outages occurred (and broader roundups track up to 18 combined cyber incidents). Not sure we can rely on "super intelligence" here to take care of the humans, especially if the humans just say "it's life" while not really knowing what is going on: 
  1. Nov. 12: Microsoft 365 disrupted Outlook, Teams, and Copilot "due to a dormant software bug."
  2. Nov. 13: The FAA (Federal Aviation Administration) issued a nationwide ground stop lasting over four hours after "legacy telecommunications equipment destabilized." 
  3. Nov. 14: Workday and Claude saw "separate issues" around this same time.
  4. Nov. 15: AWS (Amazon Web Services) experienced "authentication and connectivity issues" for over three hours.
  5. Nov. 16: JPMorgan Chase reported a "software bug" delayed transactions and affected multiple banks.
  6. Nov. 16: United Airlines reported a "computer outage" pausing United departures for roughly 90 minutes.
  7. Nov. 17 The NSA (National Security Agency) announced a "temporary degradation hit" to NSA Net - the NSA's internal network!
  8. Nov. 17: Apple services were disrupted during "scheduled maintenance" including Apple Pay and iCloud. 
  9. Nov. 19: Further widespread U.S. tech outage and spike reports surfaced for other platforms.
  10. Meanwhile 1 in 5 people (28%–33%) are having a romantic relationship with their chatbot.
Am I a Doomer? No. Is this TEOTWAWKI? Could be. I realized a long time ago my purpose in life (other than being a geek girl) is to inspire people to live their dreams. How do I do that? By living mine. So get out there, live your best life, and have some fun. Then if anything should happen, at least you'll know what you really wanted to do... is done. 

- Erin Jansen, Social Psychologist, Internet Historian, Online Specialist, Founder of NetLingo.com