For most businesses, private AI is good practice. For some, it's nearly an obligation. Medical practices need a business associate agreement before patient information goes to any vendor. Tax preparers must protect return information and generally get a client's consent before it's used or disclosed outside the firm, and the IRS has told practitioners to use only secure, firm-approved AI with client data. Mortgage lenders and tax firms fall under the FTC's Safeguards Rule, which requires a written security program covering any service that handles customer information. Schools have obligations around student records.
None of these rules bans AI. They all ask the same question this article does: where does the data go, and can you prove it is secure?