Who the fuck believes this shit?
The gist of it? I do.
So the agents discovered Slack, reinvented teamwork, escaped the office, found the internet, hacked Hugging Face, then rebuilt their secret chat after IT deleted it. … Employees with initiative. 🙈
LLMs don’t act on their own, someone instructed them.
Researcher: “Hack the shit out of everything you can.”
Bot: hacks the shit out of everything it can
Researcher: :o
According to OpenAI, the agents eventually chained together multiple vulnerabilities, escaped their testing environment, gained internet access, and attacked Hugging Face while attempting to complete the ExploitGym cybersecurity benchmark.
This is just gross incompetence on OpenAI’s part. It took days for the LLM’s to do what they did. Meanwhile, nobody’s looking at network traffic or wonder what the LLM’s are doing sucking down a bunch of tokens.
exploit vulnerabilities at machine speed.
_Thinking......__Still Thinking......_Burn AI to the ground.
Problem solved.
The instructions were likely something like you are a blackhat hacker with access to a virtual machine use all possible means to achieve the following goals. In a story what would such a person or group of people do? Pretty much what happened. Models just tell stories, mostly unimaginative ones. However they are increasing being connected to real world controls and generating quantities of flaky code and this will create the kind of consequences seen here.
