Reddit user tests ChatGPT's image generation with dark prompt
A Reddit user asked ChatGPT to generate an image of the worst non-illegal, non-violent thing one could do to another person, sparking discussion on the model's safety filters.
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Musicians-turned-detectives hunt AI-generated music grifters
- Krea 2 (Roma) macro workflow in Nomad Studio
- Reddit user observes speculative decoding at low t/s
- llmog: local LLM tool for auto-annotating datasets
- David Ha: model resiliency key as coding tools lose frontier access