Shadow AI

Shadow AI is the use of AI tools your organization has not sanctioned, usually through personal accounts and personal payment methods. It can indicate unmet demand and a sanctioned path that is missing or too slow.

Reported adoption and restrictions

Cisco's 2024 Data Privacy Benchmark Study, a double-blind survey of 2,600 privacy and security professionals across 12 geographies, reported that 48 percent of organizations admitted entering non-public company information into generative AI tools, and that 27 percent had banned the tools at least temporarily over privacy and security concerns. The same study found 63 percent had established limits on what data can be entered and 61 percent had limits on which tools can be used.

These restrictions do not identify the work people were trying to complete. If that demand remains, users may move the activity to accounts and devices the organization cannot observe.

Source: Cisco 2024 Data Privacy Benchmark Study and the accompanying release.

Failure mode S B

Three pastes in twenty days

Samsung Electronics / 2023

Samsung's semiconductor (DS) division lifted its internal ChatGPT ban on March 11, 2023; within twenty days, three employees had fed confidential material into the chatbot. Two pasted proprietary source code (a fab equipment measurement database program submitted for a bug fix, and a yield and defect identification program submitted for optimization); a third uploaded a transcribed recording of an internal meeting to generate minutes. Samsung capped prompts at 1,024 bytes as an emergency measure, then in May 2023 banned generative AI tools on company devices and networks entirely.

The skipped letter. Secure is the missing letter: this is the canonical shadow AI incident, in which governance consisted of a permission memo with no technical enforcement between engineers and a consumer service that retained inputs for training. The DS division opened access on March 11 with policy words alone; no DLP on the egress path, no prompt inspection, no data classification gate, so the confidentiality boundary held for less than twenty days. Build is the secondary gap: engineers had real jobs to do (debug code, optimize a test sequence, write minutes) and no sanctioned tool that could do them, so demand routed to the unsanctioned one. The 1,024-byte cap and the May ban were both remediation after the fact; the ban also confirms the enablement failure, since removing the tool was the only control Samsung could actually enforce.

The Economist Korea (exclusive, March 30, 2023): Samsung ChatGPT misuse surfaces as soon as the ban is lifted / Bloomberg (May 2, 2023): Samsung bans staff's AI use after spotting ChatGPT data leak / AI Incident Database, Incident 768: ChatGPT reportedly implicated in Samsung data leak

Discovery sources

Start with expense reports for AI subscriptions reimbursed to personal cards. They identify the tool, team, and payment history.
Review SSO and OAuth grants in the identity provider for third-party applications that were not approved.
Use network and DNS telemetry to measure traffic to known AI provider domains from managed devices on managed networks.
Review browser extensions that can read pages or send content to third-party services.
Run a non-punitive survey that asks which tasks people were trying to complete, then publish the approved alternatives developed in response.

Response process

After discovery, address the task that led to each tool:

Record the work the user needed to complete and the gap in the approved path.
Provide an approved option for common tasks that is usable enough for daily work.
Migrate accounts and organizational data before removing access to the old tool.
Close the unsanctioned route after the replacement is available so users are not pushed toward hidden adoption.

Personal account risk

A tool on a personal account can hold organizational data outside company control. The account may use unreviewed terms and a password the organization cannot rotate. It may also lack a supported export path.

Standard offboarding does not remove data from that account when the person leaves. Prioritize migration of personal accounts that hold organizational data.

Questions to ask your vendor, or your own team

Pull the last six months of expense reports for AI tool subscriptions. What is on the list?
Which third-party applications have OAuth grants against your identity provider?
What does someone in your organization do today when they need an AI capability you do not provide?
If a person who set up a personal AI account left tomorrow, what would happen to the data in it?
Could you tell whether someone in accounting is using the company assistant for personal work? If not, you also cannot tell what anything costs or who caused a spike; the attribution gap is the same gap.
Have you asked, without consequences attached?
For each shadow tool you find, what is the sanctioned equivalent, and how long until it exists?