Massive Bio has made its inaugural strategic investment in Rivvi. Read the announcement

← Glossary

Definition · Allowed AI at work

Shadow AI

Shadow AI is the use of AI tools at work without the organization's approval or oversight. A common example is staff pasting patient lists into a personal ChatGPT account. In healthcare it creates HIPAA risk. Consumer AI tools usually come without a Business Associate Agreement, so patient data entered into them can be an impermissible disclosure.

What shadow AI is

Shadow AI is the AI version of shadow IT. Staff adopt tools on their own because they help with the work, without security review, a contract, or a BAA. Examples include drafting patient letters in a free chatbot or summarizing a payer report in a personal account. Another is a browser extension that reads every page.

It is common in healthcare. Wolters Kluwer surveyed 518 providers and administrators, releasing results in January 2026. 40% had encountered unauthorized AI tools at work, and 17% had used them. In a Black Book survey released December 2025, 58% of frontline health-system staff used generic AI tools for work in the past 30 days. Of those users, 17% sometimes or often included identifiable patient information.

Shadow AI usually signals unmet demand, not bad intent. In Elsevier's 2025 Clinician of the Future survey, only 32% of clinicians said their institution provided adequate AI access. Bans alone tend to push use further out of sight. The fix is an approved tool with a BAA, plus a clear AI acceptable use policy. Rivvi is built to be the AI a healthcare team is allowed to use at work.

Questions

Shadow AI, answered

What is an example of shadow AI in healthcare?
A care coordinator exports a list of overdue patients and pastes it into a personal ChatGPT account to draft outreach scripts. The tool is unapproved, no BAA covers it, and the list contains PHI. Nobody in IT or compliance knows it happened, and no audit trail exists.
Why is shadow AI a risk for HIPAA?
Consumer AI plans generally do not come with a BAA. OpenAI's July 2026 HIPAA guide, for example, says ChatGPT Free, Plus, Pro, and Business are not eligible for PHI. Entering patient data into an uncovered tool can be an impermissible disclosure, and the organization has no control over retention or access.
How do you stop shadow AI?
Give staff an approved AI tool that is covered by a BAA and good enough that they want to use it. Pair it with a written acceptable use policy, short training, and a way to report mistakes without punishment. Monitoring helps, but blocking without an alternative rarely works.
Is there a free AI tool staff can use with patient data?
Yes, if it includes a BAA. Rivvi's free plan is you and two colleagues, with a HIPAA BAA. No card. No clock. Staff can chat, upload patient lists and payer reports, and analyze them inside a workspace covered for PHI. Calls, texts, the website widget, and a phone number are on Team.

Try it on your own data today.

Free to start. Most teams are using it the same day.