Please read this paper & thread about actual safety rather than the co-opted "AI safety" which is the opposite 👇🏽