Why Uploading Files to AI Tools Puts CRE Firms at Risk
It’s the last Thursday of the month, and Matilda* on your team is trying to close out common area maintenance reconciliations before the quarter ends. There’s a stack of vendor invoices to cross-reference against the CAM budget – landscaping, snow removal, the elevator maintenance contract, three different cleaning vendors. Doing it by hand will take the rest of the day. So she scans the invoices, uploads the batch to a free AI tool, and asks it to total everything by category. Ninety seconds later, she has her answer. She moves on to the next task and never thinks about it again.
Down the hall, Louie* is doing something similar. A batch of tenant leases needs review – which tenants hold a first right of refusal, which renewals are coming due in the next twelve months, what notice period each lease requires before a decision has to be communicated. It’s exactly the kind of task that used to take a paralegal the better part of a week. Uploaded to an AI tool instead, it takes an afternoon.
How BII Runoff Happens Without Anyone Realizing It
These are real stories. We found them when we went looking for BII Runoff – that phenomenon when data sluices out of the company and into the wild via an innocent-looking AI tool – the same tool that Matilda used last night to envision new decor for her living room wall.
Neither of these people did anything wrong, by their own understanding of the job. They used the fastest available tool to solve a real problem. And in both cases, information that had no business leaving the building already has.
That exposure carries real consequences – regulatory, financial, and reputational – that most organizations don’t fully account for until it’s too late, and the information that had no business leaving the building has already triggered those consequences.
What Actually Left With It
The A/P invoices uploaded to calculate CAM charges didn’t just carry line-item totals. Vendor invoices routinely include bank account numbers and ACH routing details – the information vendors provide so they can get paid electronically.
The lease files didn’t just carry renewal dates. They carried tenant financial covenants, personal guarantees, contact information, and in some cases banking details tenants use for direct rent debits.
The risk to partners, vendors, and tenants does not overshadow the internal risk. A free AI tool now has data about your… yes, your vendors, partners, and tenants – and that data is now sitting somewhere a bad actor, or even a competitor, could get to it and use it against you. And when employees reach for these tools without oversight, shadow AI and shadow IT now compound each other – creating exposure that no single policy was designed to catch.
Could that damage your business? Cause you financial losses? Create a reputation management problem? You bet it could.
What Is BII Runoff and Why CRE Teams Are Especially Exposed
We call this pattern BII Runoff – the uncontrolled leakage of Business Identifiable Information into AI tools outside an organization’s visibility or control. BII is the term federal agencies use for confidential commercial and financial information – the business equivalent of PII. It’s not a new category of data. It’s the same category that’s always mattered.
What’s new is how easily it now flows out the door, one convenient upload at a time.
When AI Uploads Expose Your Tenants’ and Vendors’ Data Too
That’s a different kind of risk than the one most AI-usage conversations focus on. It’s not just “did we protect ourselves.” It’s “did we just make someone else’s problem for them, without their knowledge or consent.”
Real-World Examples: Samsung and JPMorgan’s AI Data Incidents
This isn’t hypothetical or unique to real estate. In 2023, engineers at Samsung uploaded proprietary source code and a transcript of an internal meeting into ChatGPT while trying to solve routine problems faster – not realizing the tool could retain and train on what they’d shared. The company restricted employee access within weeks.
Around the same time, JPMorgan Chase moved to limit employee ChatGPT use after staff had used it to summarize confidential client communications and trading data. Neither company’s employees intended to leak anything. Both companies ended up managing an incident anyway.
Why an AI Policy Alone Won’t Stop BII Runoff in CRE Firms
The instinct is to write a policy: don’t upload confidential information to AI tools. It doesn’t hold, and everyone quietly knows it won’t, because the tools are genuinely useful and genuinely fast, and a rule with no approved alternative just gets quietly ignored the next time someone’s under deadline pressure. 
What Actually Closes the Gap: Visibility Plus an Approved Alternative
What actually closes the gap is knowing what’s already happening – which tools, which categories of information, whose data is involved – and then giving employees an approved way to get the same speed without the exposure.
That’s the difference between a policy that sits in a drawer and one that actually changes behavior. And the governance guardrails that prevent data exposure need to be incorporated in the corporate policy documents, hopefully long before there’s a need for them..
How ProtectMyIT’s BII Containment Works
BII Containment is how we do that: a full inventory of every AI tool in active use and what’s been shared with it, an assessment of whether those tools retain or train on what they receive, an approved-tool policy that replaces the free-for-all with an actual decision, and plain-language employee guidance – followed by ongoing management, because new tools keep showing up and the exposure doesn’t stop on its own.
The Specific CRE Risk: Finance and Leasing Documents Carry Third-Party Data
We’ve written before about the broader shift this sits inside – shadow AI operating alongside sanctioned AI inside most organizations. BII Runoff is the specific, CRE-relevant version of that pattern: the exact documents your finance and leasing teams handle every day are exactly the documents most likely to carry someone else’s confidential information along with your own.
Most organizations are surprised by what they find once they actually look. Matilda and Louie aren’t going anywhere, and neither is the shortcut they reached for – it’s simply the fastest tool available for a real job. But somewhere outside your building right now, information belonging to people who never agreed to that risk is sitting inside a system nobody at your company controls. The question isn’t whether that keeps happening. It’s whether you’re the one who finds out first.
*Don’t worry – the names have been changed to protect the… well, the guilty, in this case.
Frequently Asked Questions
What is BII Runoff and how does it happen?
BII Runoff is the uncontrolled leakage of Business Identifiable Information into AI tools outside an organization's visibility or control. BII is the term federal agencies use for confidential commercial and financial information — the business equivalent of PII. What's new is how easily it now flows out the door, one convenient upload at a time.
Why doesn't an AI usage policy alone stop employees from uploading confidential files to AI tools?
The instinct is to write a policy: don't upload confidential information to AI tools. It doesn't hold, and everyone quietly knows it won't, because the tools are genuinely useful and genuinely fast, and a rule with no approved alternative just gets quietly ignored the next time someone's under deadline pressure.
What data risks are CRE firms specifically exposed to when employees upload lease and finance documents to AI tools?
The lease files didn't just carry renewal dates. They carried tenant financial covenants, personal guarantees, contact information, and in some cases banking details tenants use for direct rent debits. The exact documents your finance and leasing teams handle every day are exactly the documents most likely to carry someone else's confidential information along with your own.
