The real AI risk is inside the labs

This blog post argues that the real risks associated with AI development are not merely related to open models, but rather the activities within AI labs, particularly regarding leaks and potential unethical use of AI technologies. The author critiques concerns expressed by Amodei, stressing the necessity for global safety standards and collaboration in the face of competitive AI advancements. They highlight the nuanced dangers AI presents, including potential benefits in medicine weighed against the risks of catastrophic misuse.