Should AI research be controlled? If yes. How?

What do you guys think? I personally think it shouldn’t be. But I can be biased, as I run Wavel AI.