Protect AI is a US AI security company that Palo Alto Networks acquired in 2025; its technology is integrated into Prisma AIRS. The platform helps organisations secure AI models, run AI red teaming and protect AI at runtime. Prisma AIRS covers model security (scanning third-party models for tampering, malicious scripts and deserialisation attacks), posture management, runtime security, red teaming and agent security. It is delivered as a managed platform, with network intercept and API intercept among the deployment modes; the software itself is not open source.
Prisma AIRS belongs to the US company Palo Alto Networks, so the CLOUD Act is relevant. For an API intercept deployment you choose a region: the Americas, Europe (Germany), India, Singapore or Japan, and that choice sets where data is processed and stored. According to the documentation, some detections, such as contextual grounding, are processed outside the chosen region, via the Netherlands or Singapore. Check this against your own scenario.