Definition
The security and availability risk that arises when multiple organisations share the same AI platform, model-serving infrastructure, or agent orchestration environment. An attack—or even unusually heavy legitimate use—by one tenant can degrade service or leak data for all other tenants.
Why it matters
Most enterprise AI platforms today are shared infrastructure. A denial-of-service attack that exploits the AI guardrail layer (as in reasoning-extension DoS attacks) or a cross-tenant data access vulnerability affects every customer on the platform simultaneously, not just the targeted organisation.