• Manana Silas
    Manana Silas Commented
    · 10 hours ago ·
    Manana Silas
    Manana Silas
    Good point on abliterated models. Security teams should treat agent tool access as a separate risk layer from the model itself. #AIsecurity