Their intent really isn't what matters.

Would you argue that its impossible for OpenAI or a similar company to accidentally develop something dangerous while they believe they are developing a safe, controlled but potentially much more powerful tool?