MIT's New Method Flags AI Models Trained on CASM Without Generating It

Hacker News points to a new method developed at MIT that can flag whether an AI model was trained on child sexual abuse material, without generating any of that imagery to run the check. The feed carries only the headline, so it doesn't include further detail on how the technique works.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton