ASCII art elicits harmful responses from 5 major AI chatbots

8 months ago 171

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

View Entire Post

Read Entire Article
  1. Homepage
  2. Technology
  3. ASCII art elicits harmful responses from 5 major AI chatbots

Related

More News From Ars Technica

Trending

1. John Fetterman
2. Karen Read
3. Florida Panthers
4. Georgia baseball
5. Malawi
6. Tony Evans
7. Rebecca Grossman
8. iOS 18 beta
9. Maren Morris
10. Nvidia stock price

Popular

What is Authentic Project Based Learning (PBL)?
4 weeks ago

Wall Street hits record high after Trump election win, as US dollar, bitcoin and Tesla shares surge – as it happened
3 weeks ago

Marathe departs ALM for The Insurer, a Reuters publication
4 weeks ago

A single grant of $123M has been finalized from $33B+ announced so far under the US CHIPS Act, which is facing criticism on both sides of the political aisle (Politico)
3 weeks ago

The NYC Marathon Is the Best Pick-Me-Up
3 weeks ago

‘South Park’ Takes on Big Pharma, Bigger Influencers
4 weeks ago
© Obiaks 2024. All rights are reserved