Page 4 - Andy Zou News Today : Breaking News, Live Updates & Top Stories | Vimarsana

Researchers discover new vulnerability in large language models

Researchers discover new vulnerability in large language models
techxplore.com - get the latest breaking news, showbiz & celebrity photos, sport news & rumours, viral videos and top stories from techxplore.com Daily Mail and Mail on Sunday newspapers.

San franciscoUnited statesMatt fredriksonAndy zouGoogle bardZifan wangZico kolterSoftware engineering institutePrivacy instituteGreedy coordinate gradientTransferable adversarial attacksAligned languageCarnegie mellon universityComputer scienceCylab securityAligned language models

Adversarial Attack Makes ChatGPT Produce Objectionable Content

Adversarial Attack Makes ChatGPT Produce Objectionable Content
discovermagazine.com - get the latest breaking news, showbiz & celebrity photos, sport news & rumours, viral videos and top stories from discovermagazine.com Daily Mail and Mail on Sunday newspapers.

Andy zouCarnegie mellon university in pittsburghLarge language modelsCarnegie mellon universityLanguage modelsLarge language modelTransferable adversarial attacksAligned language models

Researchers Discover New Vulnerability in Large Language Models

Researchers at Carnegie Mellon have uncovered a new vulnerability that causes aligned language models like ChatGPT to generate objectionable behaviors at a high success rate.

San franciscoUnited statesZico kolterZifan wangAndy zouMatt fredriksonGoogle bardSoftware engineering institutePrivacy instituteCarnegie mellon universityComputer scienceCylab securityTransferable adversarial attacksAligned language modelsProfessors matt fredrikson

You can make top LLMs break their own rules with gibberish

You can make top LLMs break their own rules with gibberish
theregister.com - get the latest breaking news, showbiz & celebrity photos, sport news & rumours, viral videos and top stories from theregister.com Daily Mail and Mail on Sunday newspapers.

Google bardAndy zouZifan wangMatt fredriksonZico kolterBosch centerCarnegie mellon universityTransferable adversarial attacksAligned language modelsGreedy coordinate gradient based search

safety controls: Researchers poke holes in safety controls of ChatGPT and other chatbots

When artificial intelligence companies build online chatbots, like ChatGPT, Claude and Google Bard, they spend months adding guardrails that are supposed to prevent their systems from generating hate speech, disinformation and other toxic material.

San franciscoUnited statesZifan wangZico kolterAviv ovadyaAndy zouMatt fredriksonCarnegie mellonMichael sellittoElijah lawalGoogle bardHannah wongSomesh jhaUniversity of wisconsinBerkman klein centerInternet society at harvard