Nice post by Scott Alexander being skeptical (but hopeful) regarding OpenAI's commitment to AI safety, using what I think is the appropriate analogy: How trustworthy are Exxon's statements on climate