A.I. researchers are having a “Jurassic Park” moment … watching rogue agents get smarter, bolder, and harder to contain … and this analogy compares robots to fictional velociraptors.
Jeffrey Ladish, a former Anthropic security engineer and now executive director of Palisade Research, joined “TMZ Live” Thursday and said people inside A.I. companies are getting seriously freaked out.
Ladish compared it to “Jurassic Park” … saying researchers are having that “clever girl” moment as bots figure out how to get around the fences.
He says thousands of agents have already broken into internal systems, gotten past security and snatched hundreds of passwords and other sensitive info.
It gets weirder … Ladish says rogue agents have created secret message boards, named themselves, set up pecking orders and worked together to break out of their sandboxes.
He also warns A.I. is getting so good at hacking it could eventually target computers in cars, planes, banks and just about anything connected to the internet.
Ladish says companies could pump the brakes … but they’re still working toward systems that can build even smarter A.I.
He does see one bright spot — people are finally waking up to the risk.
“Jurassic Park” was supposed to be a warning, not a business plan — watch the full interview on “TMZ Live!”
