In the linked HN comment from a few years ago, an Anthropic employee said it was some fraud detection system.

It could be that OP is unlucky, and some of his metadata (or perhaps payment information) matches some patterns for stolen-CCs/fraud/chargebacks?

I've also found Claude.ai to be very suspicious of less mainstream browsers (e.g. Pale Moon), unfortunately.