Anthropic is working on a system to prove AI models’ identity

Called FlashProve, the system would let anyone check that an AI response really came from the model it claims to be from.

The top 3

Mechanism of Cryptographic Verification: FlashProve relies on 'provable inference,' where AI labs generate and publish a 'fingerprint' for each model. This mathematical guarantee allows users to confirm the origin of an output through local verification, akin to cryptographic email signatures, without requiring access to the private, proprietary weights of the model.
Use Cases and Limitations: The system is designed to prevent model substitution, such as labs secretly swapping high-end models for cheaper ones, and to verify third-party claims about an AI's involvement in a task. However, the current release is an open-source research demonstration intended for further community development rather than a ready-to-deploy commercial product.