Quiz: Will It Leak My Data? | Learn

1. What is the key data difference between consumer and enterprise versions of AI tools like ChatGPT?

Enterprise versions always use smarter models than consumer versionsConsumer tiers may train on your inputs by default, while enterprise and API terms typically include contractual no-training commitmentsConsumer versions cannot connect to the internetEnterprise versions always run on your own hardware

Same brand, opposite data postures. The training terms, spelled out in the contract, are the difference that matters for your business data.

2. In the three-tier routing model, how is business-sensitive data such as CRM records and financials handled?

Sent to frontier models, because the best capability should winBlocked from all software, including internal toolsStored locally and queried in place, with only derived summaries ever reaching an outside APIEncrypted, then uploaded to whichever tool needs it

The middle tier keeps the raw data in local storage. Agents query it where it sits, and only summaries derived from it ever leave.

3. The lesson describes an agent being refused access to the credential store by automatic policy. Why is that refusal presented as the system working?

It proves the agent was malfunctioning and needed replacementBoundaries enforced in configuration hold without relying on anyone's memory, and a system that can refuse its own agent can refuse a compromised oneIt shows credentials should never be stored in digital formIt demonstrates that agents cannot be trusted with any task at all

Deny-by-default access is the contractor-and-keycard principle in action. The refusal proves the boundary is real rather than aspirational.

Back to lesson Next lesson →