[nohat]

I feel like Scott Alexander's thoughts on openai when it was first announced are still relevant now. I can think of strategic reasons for what openai has done, but I also feel like their tactics and 'democratization of AI' credo focus more on the short term economic and political dangers of a monopoly or oligopoly on AI than existential threat. In fact it probably increases existential risk, seeing as a competitive race (providing no time for caution) to be the first to finish GAI is surely the most plausible and possibly the most dangerous scenario.

http://slatestarcodex.com/2015/12/17/should-ai-be-open/