Claude Mythos: The Model Anthropic Built and Decided Not to Sell

Anthropic built an AI model called Claude Mythos that discovered a bug which automated testing had missed across five million runs, and the company chose not to release it, according to a Towards AI report. The model's successor shipped three days before the US government took action, the report states.

It found a bug that automa‍te‍d test‌i‍ng h⁠a‌d run‍ pas​t fi‍ve mi‍llion tim​es. Then, three day​s after​ its success‍or shipp⁠ed, the US… Continue reading on Towards AI »