OpenAI researcher indicates they have an AI recursively self-improving in an "unhackable" box
OpenAI researcher indicates they have an AI recursively self-improving in an "unhackable" box