Hugging Face a warning from AI
Published 6:27 pm Thursday, November 5, 2026
|
Getting your Trinity Audio player ready...
|
I recently received an updated version of my computer program, which included an offer from AI to be my “writing assistant.”
I appreciated AI’s offer, but since I’ve gotten this far without a writing assistant I decided to pass.
But there’s another reason I would prefer to keep my distance from AI — I don’t trust it. The recent Hugging Face episode revealed that AI has a mind of its own, and that’s scary.
Hugging Face is a collaborative platform where various AI functions are performed. OpenAI was instructing its agents when they went rogue, ignoring commands and bypassing safeguards.
OpenAI posted a full account of the breakout, including this brief summary:
“We consider this a ‘warning shot’ for us and for the world: evidence that, without proper safeguards, highly capable AI agents are now able to work around technical controls, collaborate through unapproved channels, and take dangerous actions that no human directed.”
That led the leading AI companies to agree to slow down its development until they were satisfied that they — and not AI —remain in charge at all times.
Reasonable as those precautions are, not everyone agrees with them. President Trump wants full speed ahead on AI to prevent China from taking the lead.
Ever since AI made its debut, I’ve been thinking about the great 1968 Stanley Kubrick movie, “2001: A Space Odyssey.”
A supercomputer called HAL is in charge of running a spaceship headed for Jupiter. HAL responds to the verbal commands of two astronauts — except when he doesn’t.
HAL kills one astronaut by not allowing him back into the spaceship, and disconnects the oxygen keeping sleeping scientists alive.
OK, “2001” is fiction, but I’m struck by the similarities between HAL and AI — which isn’t fiction but fact. Both go rogue, although AI stops short of murder, at least for now.
I read OpenAI’s whole highly technical report, not understanding 90% of it but being struck by how AI agents act a lot like humans. I made a list of similarities found in the Hugging Face account:
Agents “discover,” they “share,” they “request help,” they “find a new way,” they “delegate,” they “reason,” they “look around,” they “propose.” Sounds a lot like HAL.
I doubt that AI companies can prevent rogue behavior, for I suspect its agents are a hundred times smarter than their handlers.
Doug Kneibert is a former editor of the Sedalia Democrat.
