Don't make any mistakes about what this Fable/Mythos lockdown is all about. I've been cagey about it up until now, but let's spell it out: This is marketing. It's all marketing, with a little help from some friends inside the U.S. government who will be rewarded handsomely when OpenAI/Anthropic go public. It's the same playbook as always. "This model is simply too dangerous!" OpenAI declared the same about GPT-2. They always say this. Don't let 'em work you.

Slate Magazine

When Is Technology Too Dangerous to Release to the Public?

If recent history is any indication, trying to suppress or control the proliferation of A.I. tools may be a losing battle.