Practical AI cover image

Vector databases (beyond the hype)

Practical AI

00:00

New Automated Adversarial Attacks on Language Models

New research has uncovered automated, effective attacks on large language models like chat GPT, bard, and clawed. These attacks involve constructing specific character sequences that, when added to a user query, can trick the system into following harmful commands. Unlike previous jailbreak attempts, these attacks are automated and potentially unpatchable by LLM providers.

Play episode from 15:58
Transcript

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app