That's a nonsensical website. A third party was responsible for both OpenAI's and Anthropic's model escapes. The difference is that Anthropic was warning that this might happen, and OpenAI was putting their head in the sand.