One of Google’s recent Gemini AI models scores worse on safety https://t.co/UNyRxmUD6a https://t.co/5glzL0mU1M on https://twitter.com/AcerboLivio/status/1918402410503942532