AI model audits made tamper-proof with cryptographic receipts
A Python library now generates tamper-proof cryptographic receipts for AI model outputs, proving what was generated, when, and by which model—even when invisible watermarks are embedded in text.