Some workers argued that OpenAI’s controls were so lax that outsiders, including foreign governments, could obtain sensitive information about the AI lab’s technology with relative ease, the people said