Prompt Security covers the risks organisations inherit when staff and applications start using generative models: prompt injection, sensitive data leaving through a chat window, unsanctioned tool use, and attacks against AI agents. It sits between users or applications and the models, inspecting traffic in both directions. The company was acquired by SentinelOne and its technology now sits within that vendor’s platform, which buyers should factor into procurement.
Shadow AI is the problem most security teams underestimate. Staff paste customer data into consumer assistants long before a policy exists, and visibility over that traffic is the first control worth having.
Alongside the commercial product, the team publishes open-source tooling and research, which is a reasonable way to assess their depth before any purchasing conversation. Pricing is enterprise and quoted after scoping.
