“Jan Leike, former head of superalignment at OpenAI, wrote to the board with whistleblower-style material stating the company was going 'off the rails' on its safety mission.”