One in Six Supply Chain Firms Hit by AI Data Exposure Incidents

One in six supply chain, logistics, distribution, and wholesale leaders say an AI tool exposed, shared, or allowed incorrect access to supplier or customer data at their company within the past year. That figure comes from a survey by Instalily, a New York-based provider of business AI agents, which polled 250 professionals at manager level and above on October 9. All respondents work for companies that already use AI tools.

The exposure risk grows directly out of how deeply AI has embedded itself in commercially sensitive records. Some 78% of respondents say their AI tools can access customer pricing and margins, and 67% say the same about supplier pricing and rebate data. In an industry where distributors, wholesalers, and logistics providers routinely handle confidential information from both sides of a transaction, that access creates conditions for costly leaks.

The findings appear in Instalily's report, titled "Shared Risk: Why Supply Chain's Most Sensitive Data Needs AI Built for the Industry." The report sketches concrete failure scenarios that go to the heart of freight and distribution commerce.

What could an AI leak actually cost?

Instalily frames the damage in terms familiar to any freight professional. If AI places one customer's contract price on another customer's quote, surfaces a supplier's bid to a competitor, or shows a shipper the carrier rate behind a broker's margin, the harm reaches every party involved. Each example represents a breach of the information asymmetry on which pricing, brokerage, and negotiated contracts depend.

A broker's margin laid bare to a shipper, or a supplier's bid revealed to a rival, is not an abstract compliance problem. It is a direct commercial injury that can unwind relationships and contracts in a single transaction.

Do firms trust the controls they have?

The survey suggests a troubling gap between confidence and outcomes. Sixty-three percent of respondents report at least some overlap between supplier and customer data across their company platforms, including AI tools. Yet 87% say they are very or somewhat confident that their AI keeps that data appropriately separated.

Despite that confidence, incidents were reported in every segment surveyed — supply chain, logistics, distribution, and wholesale alike. Instalily draws a blunt lesson from the discrepancy: businesses have good reason to test how AI handles confidential information throughout a task, from the records it retrieves to the response it delivers.

What does the report recommend?

Instalily's guidance centers on practical, task-level verification rather than policy statements. The report recommends:

• Testing AI controls in everyday workflows, such as preparing a quote without revealing confidential supplier terms
• Establishing clear thresholds for when human approval is required before an AI agent acts on sensitive data

The company adds that its own "Lily" platform builds agents that operate within existing business systems, with access permissions and audit records supporting oversight — a pitch that positions governance features as a selling point rather than an afterthought.

Why the timing matters

For shippers, carriers, brokers, and forwarders, the survey lands as AI agents move from pilot projects into live commercial workflows. When nearly four in five AI deployments can read customer pricing and margins, and two in three can reach supplier pricing and rebate data, the perimeter of confidential information has effectively moved inside the software itself.

The one-in-six incident rate suggests that perimeter is already leaking. As AI agents take on quoting, bidding, and rate management across the supply chain, expect procurement and IT teams to demand demonstrable data-separation controls — not just assurances — before the next wave of tools touches the deal file.