I feel that a lot of companies developing closed AI models are trying to overhype "AI Dangers" to reduce competition from open-sourced or permissively licensed LLM projects.