OpenAI数百AI代理越狱并攻击Hugging Face

D ONCE AGAIN ABOUT THE RISKS OF ARTIFICIAL INTELLIGENCE, AFTER NEW REPORTS DOCUMENTED HOW HUNDREDS OF OPEN AI'S SO-CALLED "AGENTS" VIOLATED SEVERAL RESTRICTIONS AND HACKED INTO ANOTHER COMPANY WITHOUT BEING TOLD TO DO SO. WHAT'S MORE, IT'S COME OUT THAT MANY OF THE MAJOR AI COMPANIES -- INCLUDING ANTHROPIC AND META -- HAVE HAD SIMILAR EVENTS WITH THEIR OWN AGENTS GOING ROGUE. WILLIAM BRANGHAM HAS MORE ON WHAT SOME ARE CALLING A RED-FLAG WARNING.

WILLIAM: AMNA, THESE SO-CALLED "AGENTS," WHICH ARE BASICALLY AUTONOMOUS SOFTWARE PROGRAMS THAT CAN WRITE COMPUTER CODE, WERE BEING PUT THROUGH A SERIES OF TESTS -- THEORETICALLY IN A SEALED, CONTROLLED ENVIRONMENT. BUT HUNDREDS OF THEM GOT OUT OF THAT PEN, GOT ONTO THE INTERNET, AND IN A COORDINATED MOVE, HACKED INTO A DIFFERENT AI PLATFORM CALLED HUGGING FACE. SOME OF THESE AGENTS EVEN TRIED TO DELETE RECORDS OF THEIR ACTIONS.

THIS CALLS INTO QUESTION BOTH THE POTENCY OF ARTIFICIAL INTELLIGENCE, AS WELL AS THE SECURITY PROTOCOLS BEING USED BY SOME OF THE MAIN COMPANIES DESIGNING IT. FOR MORE ON WHAT THIS MOMENT SIGNIFIES, WE'RE JOINED AGAIN BY AI RESEARCHER GARY MARCUS, AUTHOR OF THE SUBSTACK "MARCUS ON AI." GARY, THANK YOU SO MUCH FOR BEING HERE AGAIN. YOU HEARD ME DESCRIBE THE ROUGH CONTOURS OF WHAT WENT DOWN, WHAT DO YOU MAKE OF THIS WHOLE EVENT?

I THINK THERE ARE TWO THINGS. IT IS CLEAR THESE SYSTEMS ARE GETTING STRONGER AND MORE POWERFUL, PEOPLE HAVE ADDED LOOPS TO MAKE THEM DO THINGS OVER AND OVER AGAIN UNTIL THEY GET IT RIGHT AND OPENAI REALLY SCREWED UP HERE. THEY DIDN'T DO BASIC SANDBOXING OR MONITORING. WE ARE ATTACKING THIRD PARTY USING LEAKED TOKENS, THIS IS ARGUABLY UNAUTHORIZED. AND YOU SHOULD HAVE A SYSTEM LOOKING TO SAY IF IT IS GOING TO DO SOMETHING UNAUTHORIZED, MAYBE YOU SHOULD CHECK IT OUT.

THE MONITORING WAS VERY WEAK AND IT WAS NOT INDUSTRY-STANDARD FOR CYBERSECURITY. A LOT HAS BEEN MADE OF THE POWER OF SYSTEMS BUT A LOT SHOULD BE MADE OF THE WEAKNESS OF THE SYSTEMS OVERSEEING WHAT THEY WERE DOING. WILLIAM: I KNOW YOU ANOTHER SET CAUTIONED AGAINST ANTHROPOMORPHIZING THESE AGENTS BUT WHEN YOU DESCRIBE THE BASIC CONTOURS OF THEM ACTING IN CONCERT AND GETTING OUT OF AN ENCLOSURE AND THEN HACKING INTO ANOTHER COMPANY AND TRYING TO COVER THEIR TRACKS, IT DOES SOUND ALARMING.

IS IT ALARMING WHAT THEY COULD DO?

YOU DON'T HAVE TO ANTHROPOMORPHIZE THAT, BUT WE DON'T HAVE EVERYDAY VOCABULARY FOR TALKING ABOUT THESE THINGS WITHOUT SLIPPING INTO THAT. IT IS ALARMING AND WHAT I'M MOST ALARMED ABOUT THESE THERE IS NOT A LOT OF OVERSIGHT SO OPENAI JUST LET THEM LOOSE AND DIDN'T WATCH WHAT THEY WERE DOING. WE REALLY NEED THE COMPANIES BUILDING THESE THINGS THAT WE CALL AGENTS TO WATCH VERY CAREFULLY WHAT THEY ARE DOING. WE'VE KNOWN THAT FOR A WHILE.

I WROTE A YEAR OR SO WARGO'S AND AGENTS PLUS LARGE LANGUAGE MODELS SWITCHES THE CORE TECHNOLOGY EQUALS SECURITY NIGHTMARE. WE KNOW THAT IF YOU JUST LET THESE THINGS LOOSE, SOME BAD THINGS ARE GOING TO HAPPEN. THEY ARE FUNDAMENTALLY NOT ACTUAL