Anthropic unveils FlashProve for provable LLM inference
According to Anthropic, new research introduces FlashProve, an open source cryptographic method to verify a response came from a specific LLM, improving efficiency over prior open source approaches and enhancing chat integrity and model evaluation.
SourceAnalysis
Anthropic announced new research on provable inference for large language models on November 3, 2026.
The company introduced FlashProve, a system built on established cryptographic techniques that lets developers verify whether a given response came from a specific AI model, and is substantially more efficient than prior open-source methods according to Anthropic.
These proofs can make it harder for attackers to hijack a chat and enable safety evaluators to confirm that the model under test matches the one served to customers, with the code open-sourced by Anthropic.
Anthropic
@AnthropicAIWe're an AI safety and research company that builds reliable, interpretable, and steerable AI systems.