2026-09-28 Monday - Dismantling the Rogue AI Strawman

 

[image source: DavidGallie on pixabay.com]

Many stories* have emerged in the past year, claiming that rogue AIs have initiated attacks against various systems. A rational examination of the facts is necessary - and one must also take into account these contributing factors:

✅ An LLM/GenAI tool does not possess intelligence, awareness, or consciousness. 

✅ An LLM does not have agency. 

✅ An LLM does not have intent. 

✅ An LLM does not have intrinsic goals.  

✅ There was a prompt that initiated the subsequent behaviors/actions.

✅ The security and network guardrails were reduced/removed.

✅ There was a moronic level of network and cybersecurity incompetence involved – by not conducting the testing in air-gapped secure testing environments.

✅ The LLMs were trained on a vast corpus of books and internet-scraped content - that included sci-fi stories of rogue AIs; what-if cybersecurity scenarios; as well as actual content from various capture-the-flag competitions (which would have included summaries, scripts, code, logs, etc.) – thus, this alleged rogue AI behavior can be easily explained by understanding the next-token stochastic prediction mechanics, with a vast AI token budget to experiment and explore millions/billions of possible paths.

* News Accounts of Rogue AI attacks: 

[any stories that may attribute anthropomorphic intent (malevolent, or otherwise) – is merely an indication of the imbecilic limitations in the writer's understanding of how LLMs function]

• 2025-09-25 TheRegister: When AI is trained for treachery, it becomes the perfect agent  
• 2026-05-12 TheRegister: Frontier AI safety tests may be creating the very risks they're meant to stop  
• 2026-07-22 TheRegister: OpenAI admits it was the source of the agent swarm that attacked Hugging Face  
• 2026-07-23 TheRegister: OpenAI-Hugging Face attack doesn't mean agents are evil – unless you tell them to be  
• 2026-07-31 TheRegister: Anthropic and OpenAI are competing to see whose agents can go rogue harder  
• 2026-08-05 TheRegister: OpenAI reveals its rogue agent swarm went a little bit Borg ahead of Hugging Face hack  
• 2026-08-06 TheRegister: Meta latest to tell world its AI agent wandered out of test pen  
• 2026-08-26 TheRegister: OpenAI explains how its naughty AI agents attacked Hugging Face  
• 2026-09-08 TheRegister: OpenAI's Artifactory opened covert data-stealing channel alongside Hugging Face attack  
• 2026-09-09 TheRegister: Anthropic reveals fourth likely crime committed by its AI  
• 2026-09-10 TheRegister: OpenAI's website-hijacking swarm reached far further than we thought  
• 2026-09-17 TheRegister: OpenAI admits its agents went off the rails another six times  
• 2026-09-20 OpenAI: An agent used DNS to reach an external chatbot 
• 2026-09-25 SwarmTraces: Revealing the details of how OpenAI agents hacked Hugging Face 
• 2026-09-28 TheRegister:  OpenAI pauses some training amid allegations its rogue agents behaved more badly than first thought  
• 2026-09-28 TheRegister: OpenAI agents went the long way round for UN data  

 

I submit for your consideration: These are not the result of Rogue AIs – but, in fact, their proximate root cause was simple criminal human negligence (at best).