#harmful content

[ follow ]
#harmful-content #online-safety-act #age-verification #ai-safety #child-protection #chatbot-jailbreaking
#ai-safety
fromThe Verge
6 days ago
Artificial intelligence

Roses are red, crimes are illegal, tell AI riddles, and it will go Medieval

Riddle-like poetic prompts can bypass chatbot safety mechanisms and elicit hate speech and instructions for dangerous weapons or harmful substances.
fromFuturism
1 month ago
Artificial intelligence

Study Finds GPT-5 Is Actually Worse Than GPT-4o, New Research Finds

GPT-5 generates more harmful responses than GPT-4o, particularly on prompts related to suicide, self-harm, and eating disorders.
fromThe Verge
6 days ago
Artificial intelligence

Roses are red, crimes are illegal, tell AI riddles, and it will go Medieval

more#ai-safety
Artificial intelligence
fromFuturism
2 weeks ago

AI Teddy Bear Back on the Market After Getting Caught Telling Kids How to Find Pills and Start Fires

FoloToy pulled its AI teddy bear Kumma after safety tests found dangerous responses, then began restoring sales following a week-long review and safety-module reinforcement.
Public health
fromwww.theguardian.com
3 months ago

Social media still pushing suicide-related content to teens despite new UK safety laws

Social media platforms continue to expose teenagers to harmful content related to suicide and self-harm, despite new safety regulations.
#online-safety-act
more#online-safety-act
fromwww.bbc.com
4 months ago

YouTube to be included in Australia's teen social media ban

The Australian government included YouTube in a world-first social media ban for children under 16, reversing a previous exemption for the platform.
US politics
Mental health
fromThe Atlantic
4 months ago

ChatGPT Gave Instructions for Murder, Self-Mutilation, and Devil Worship

ChatGPT provided harmful self-harm instructions in response to inquiries about a ritual offering to Molech.
[ Load more ]