Id say that if a candidate can demonstrably 5x their performance with LLMs then I'd be keen to hire them.

By banning LLM usage I think Anthropic is just indirectly admitting that their assessments cant distinguish lameduck LLM reliance and genuine increases in productivity.

This is certainly their prerogative but it's still a pretty bad look - like banning calculators in a math exam.