Anthropic releases FlashProve to verify model identity
Understand the story
On November 4, 2026, TechCrunch reported that Anthropic released FlashProve, an open-source system for verifying the identity of AI models. Based on cryptographic proofs, the system can verify that a given output truly comes from a specific model—a concept known as "provable inference." It works by requiring model labs to publish model fingerprints that do not reveal model weights, preventing scenarios where a cheaper model is swapped in for an advanced one or services are offered under the name of an advanced model, and it supports verification completed locally on devices. FlashProve's technical specifications and code have been open-sourced on GitHub. Anthropic said the project is currently only an intermediate research milestone with no deployment plans. So far this is the only report on the matter, with no further developments.
AI overview of reporting · Updated 2 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- TechCrunch · AIAnthropic is working on a system to prove AI models’ identity
Anthropic released FlashProve, an open-source system that uses cryptographic proofs to verify that a given output truly came from a specific model, known as provable inference. The system requires labs to publish a model fingerprint that does not leak the weights, preventing swaps to cheaper models or impersonation of premium models, and supports verification on local devices; the technical specification and code are open-sourced on GitHub, and it remains a mid-stage research effort with no deployment plans yet.
Story heat trend
Not enough continuous observations to show a trend yet.