Major security weaknesses found in leading open AI models

5 points | by pseudolus 8 hours ago