acm-header
Sign In

Communications of the ACM

ACM News

Researchers, Scared by Their Own Work, Hold Back 'Deepfakes for Text' AI


View as: Print Mobile App Share:
Metropolis

This is fine.

OpenAI, a non-profit research company investigating "the path to safe artificial intelligence," has developed a machine learning system called Generative Pre-trained Transformer-2 (GPT-2 ), capable of generating text based on brief writing prompts. The result comes so close to mimicking human writing that it could potentially be used for "deepfake" content.

From Ars Technica
View Full Article


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account