One in six supply-chain professionals surveyed said their company had experienced customer or supplier information being exposed through an AI tool in the past year. The findings point to a practical tension: businesses are giving AI access to sensitive commercial data while partners increasingly ask how that information is protected.
Watch Desk analysis
What happened
InstaLILY, which develops AI agents for distribution, commissioned a Pollfish survey of 250 managers and executives at companies using AI in their daily operations. FreightWaves reports that one in six respondents said their company had experienced an incident involving supplier or customer information exposed, shared or accessed incorrectly through an AI tool.
The survey also found that 78% of respondents said their company’s AI tools could access customer pricing and margins, while 67% said they could access supplier pricing and rebate information. These are respondents’ reports, not independently verified exposure incidents. The survey did not assess how serious the reported incidents were.
Why it matters
Pricing, margins and rebate terms can be commercially sensitive, and an exposure could strain relationships between trading partners. The survey suggests those relationships are already part of purchasing decisions: 63% said their company had been asked by a supplier or customer how it protects their information, and 30% said such enquiries had influenced a decision involving a software provider, vendor or AI tool.
There is a notable gap between access and confidence. While 87% said they were very or somewhat confident that their AI tools kept supplier and customer information appropriately separated, 63% also acknowledged some overlap between those data across company platforms, including AI tools. The figures describe this survey’s respondents, not the wider industry as a whole.
Our read
This is a warning signal, not an audited breach tally. The survey was commissioned by an AI-agent developer, relied on respondents’ knowledge and experiences, and did not grade incident severity. Still, the practical question is clear: can a tool prepare a customer quote without pulling in another customer’s price or a supplier’s confidential terms?
InstaLILY’s report recommends testing AI tools against routine business scenarios and deciding when a person must approve an action. That is a more useful starting point than asking whether a vendor’s security claims sound reassuring. Reassurance is plentiful; a clean test result is better.
What to watch
- Whether companies independently investigate and disclose AI-related data exposures.
- What access controls and data-separation tests vendors can demonstrate.
- Whether customers make those controls a condition of choosing AI suppliers.
Discussion spark: Should companies have to prove that AI tools keep customer and supplier data separate before deployment, or is existing vendor assurance enough?
Sources and evidence
- Source update (10 November 2026, 22:15 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.