Article

Ars Technica on 2024-03-16 01:17

Researchers use ASCII art to elicit harmful responses from 5 major AI chatbots

LLMs are trained to block harmful responses. Old-school images can override those rules.

Llama

Related news

Chatbots’ inaccurate, misleading responses about U.S. elections threaten to keep voters from polls Seattle Times
New Jersey , Nevada , United States , France , Llama
TAI chatbots provided harmful eating disorder content: report thehill.com
The Hill
Microsoft Investigates Harmful Chatbot Responses—The Latest Chatbot Blunder From Top AI Companies Forbes
Two versions of the trolley problem elicit similar responses everywhere Ars Technica
Jailbreaking LLMs with ASCII Art schneier.com
On Chatbots TechCrunch
Wall Street
AI chatbots ‘lack safeguards to prevent spread of health disinformation’Shropshire Star
Bletchley Park , Llama , Australia , Adelaide
Disinformation Researchers Raise Alarms About A.I. Chatbots The New York Times
UNM researchers study PTSD responses, seek volunteers KOB 4
Don’t Trust the Chatbots The Atlantic
Russia
Why do chatbots suck?TechCrunch
Are Chatbots Useful?Forbes
Meta launches consumer AI chatbots with celebrity avatars in its social apps Ars Technica
Brû , Llama
Google and Microsoft’s AI Chatbots Refuse to Say Who Won the 2020 US Election Wired.com
Germany , Switzerland , Russia , Llama , Washington
DOOM in ASCII mode is equal parts fun and frustrating thenextweb.com
SBaidu teases Ernie 4 large language model after China approves ChatGPT-style chatbots for public use scmp.com
Asia , Llama , Beijing , China
Chatbots' inaccurate, misleading responses about U.S. elections threaten to keep voters from polls Star Tribune
California , New Hampshire , New Jersey , Nevada , United States