Anthropic’s FlashProve proves which AI model produced a response without revealing its weights.

How they built it matters:

→ Post-quantum security
→ Hash-based STARK proofs
→ Battle-tested cryptography already used by Starknet
→ Formal verification in Lean, building on earlier work inc Cairo + Stwo

@StarkWareLtd co-founder Alessandro Chiesa is among the researchers.

https://www.anthropic.com/research/provable-inference