Class action suit against AI makers over deepfake child sexual abuse material expands
Page 9/10

A more recent version of Stability AI's model was frequently used to generate NSFW content and generated a higher proportion of such content compared with outputs from Stable Diffusion Version 2, according to a research paper by scholars from the University of Wisconsin-Madison, the Massachusetts Institute of Technology and other institutions, which has not yet been peer-reviewed. Stable Diffusion Version 2 was also much less popular than the Stability AI models that came before and after it, the researchers found.

"When they realized that nobody wanted to use their model when they put those guardrails in place, they rolled those guardrails right back so that everybody would want to use their model again," Martin said. "They knew exactly how to rein this behavior in and they chose not to for profit."

Prev
Next
Back to headlines
News menu
Home