Anthropic develops FlashProve, a model identity verification system
What happened
TechCrunch reported on November 4 that Anthropic is developing FlashProve, an open-source system that uses cryptography to check whether a given output came from the AI model it claims to use. Labs would publish model fingerprints that don't leak weights, so a swapped model can be caught, and checks can run on local devices. The spec and some working code are on GitHub; research continues and no lab has deployment plans.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- TechCrunch · AIAnthropic is working on a system to prove AI models’ identity
Anthropic 正在研发开源系统 FlashProve，通过密码学核验特定输出是否来自声称使用的 AI 模型。该系统要求实验室发布不泄露模型权重的模型指纹，旨在防范暗中替换模型，并支持在本地设备上验证。研究团队已在 GitHub 开源详细技术规范和部分可运行代码，研究仍在进行，Anthropic 或其他实验室均无部署计划。
Heat over time
Heat now 9·Comparable peak 10（Nov 4 07:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.