max-tegmark

The trap Anthropic built for itself

Anthropic, OpenAI, Google DeepMind and others have long promised to govern themselves responsibly. Now, in the absence of rules, there's not a lot to protect them.