Anthropic says Flashprove lets developers verify if a response came from a specific AI model

Provenance verification is becoming a competitive layer in the model market, not just a compliance checkbox: if a developer can confirm which model produced a given output, model identity starts to function as a verifiable attribute rather than a vendor claim. That matters most for platforms routing between multiple models and for anyone auditing AI-generated content. The open question is whether Flashprove's approach is robust enough to hold up outside Anthropic's own stack, since no technical details or release date were disclosed.