OpenAI vs Anthropic vs DeepMind: talent movement data says Anthropic is the weird one

I was looking at public career-history movement between AI companies, and Anthropic looks different from the others. Some direct movement counts I found: OpenAI -> Anthropic: 88 Anthropic -> OpenAI: 29 Google DeepMind -> Anthropic: 69 Anthropic -> Google DeepMind: 9 Google -> OpenAI: 1,448 Google -> Anthropic: 738 Meta -> OpenAI: 846 Meta -> Anthropic: 247 The interesting part is not just that Anthropic pulls from OpenAI/DeepMind. It is that the reverse direction looks much weaker. One possible interpretation: Anthropic is currently acting like a destination company for frontier AI talent, while OpenAI/DeepMind/Google/Meta are larger talent sources. Big caveat: this is based on public career-history movement data, not compensation, culture, research quality, or whether people are happy there. It also probably undercounts people who do not update profiles. Still, the directionality is pretty interesting. What do people think explains this? Better mission alignment? More upside? Better research environment? OpenAI churn? Just LinkedIn/data artifact? Methodology/source is here if anyone wants to test other company pairs: https://www.talentflow.fyi/methodology