A parks chatbot project does not need tax-lien datasets. Which data-access control best applies?

Blocking the parks chatbot project from reading tax-lien datasets it does not need is the data-access control that applies need-to-know for municipal AI. AI training and retrieval access should be scoped to project purpose. A parks chatbot answering trail hours, facility schedules, or park rules has no legitimate need for tax-lien records; denying that corpus prevents unnecessary exposure of sensitive financial and property-related data through an unrelated public-facing assistant.

Blocking the parks chatbot project from writing tax-lien datasets it does not need fails because it stops modification, not exposure; the bot could still read and retrieve lien records it has no legitimate need for, so need-to-know requires blocking read access to the tax-lien corpus entirely. Auditing the parks chatbot project for reading tax-lien datasets it does not need fails because it is detective rather than preventive; the bot could still read and expose lien records through an unrelated assistant, so need-to-know requires blocking read access up front, not merely logging it afterward. Throttling the parks chatbot project when reading tax-lien datasets it does not need fails because it still allows access at a reduced rate; the bot could still retrieve and expose lien records it has no legitimate need for, so need-to-know requires blocking read access entirely, not just slowing it.

Exam caveat: Data access for AI projects is need-to-know. If the stem says the project does not need a dataset, the correct control is to block reading that dataset—not to block only writing it or to merely audit or throttle reads.

Operational check: Review parks chatbot data scopes, deny tax-lien and other finance corpora, attempt a retrieval that would pull lien fields (fails), confirm parks content still retrieves, and verify classification labels remain enabled for municipal tables.