Shadow AI
Employees using unauthorized AI tools to get work done.
"Your real AI strategy is whatever your team's actually using."
What is Shadow AI? Employees using unauthorized AI tools to get work done. Governance gap and competitive intelligence loss.
Personal ChatGPT, Claude, and Copilot subscriptions across the org.
Governance gap and competitive intelligence loss.
Sanction great tools. Provide better internal alternatives. Train, don't punish.
Pick what to reward the model for.
Employees using unauthorized AI tools to get work done. In the wild: Personal ChatGPT, Claude, and Copilot subscriptions across the org.
Pick a lever. There are no neutral ones — every incentive funds a behavior somewhere.
Pick a reaction to Shadow AI
One tap. We'll point you at the most useful next surface based on how this hits.
The full taxonomy entry
Every concept in the Atlas uses the same structure — so Shadow AI can be compared, recombined, and cited like an element on a periodic table.
- Business
- Leadership
- Government
- Healthcare
- Education
- Sales
- Marketing
- AI
- Negotiation
- Media
- Public Policy
- Relationships
- Where in our org would Shadow AI most often show up unnoticed?
- Which metric, ritual, or contract clause quietly rewards Shadow AI?
- If we removed every payoff for Shadow AI, what behavior would replace it?
- Who benefits when Shadow AI persists — and who pays the cost?
- People defend the status quo using the language of shadow ai.
- Decisions cluster around the easiest narrative rather than the strongest evidence.
- New data changes the slide deck but not the decision.
- Anyone naming the pattern is treated as the problem.
Every Atlas entry is a node in a knowledge graph. See the related rail below to follow the connections.
See Shadow AI through 3 lenses
Each layer of the Incentives OS reframes this concept with its own thinkers, vocabulary, and diagnostic question.
Do you actually know Shadow AI?
Three quick questions. Result is saved into your review streak — come back when the term is due to lock it in.
Which best describes Shadow AI?
Worked example, counter-example & concept map
On-demand AI analysis grounded in the Lab's research. Cached on your device after first run.
Your nervous system has a region for this.
When you encounter Shadow AI, your dopamine system is tracking the gap between what you expected and what you got — and that gap is what's driving the next move, not the reward itself.
Wanting, anticipation, prediction error, motivational salience. Predictable rewards stop motivating. The phone buzz fires dopamine; the message itself rarely does.
See Dopamine in the Brain Atlas →Picked for you, from the Atlas
Ranked by shared learning paths, overlapping chips, and what you've saved.
Innovators → early adopters → majority → laggards, AI-specific.
Discounting algorithmic advice even when superior.
AI bolted onto existing workflows to look forward-leaning.
Redesigning roles around AI capability.
AI pilots that succeed and never scale.
Pre-deployment analysis of who loses what.
Send the card, not just the link
A pre-rendered social card with the title, eyebrow, and URL. Copy the link, post it anywhere, or download the SVG for slides.
More definitions to follow
Every term in the Atlas connects to a dozen others. Pick any of these and see where it takes you.
We accept vague, general statements as personally meaningful.
Who do you count as fully human, deserving consideration?
Groups sometimes behave as foolishly as the worst individuals.
When the limbic system overrides higher cognition and behavior follows ancient circuits, not chosen values.
Believing abilities are static and cannot be developed.
Tribal identity markers reward in-group loyalty over coalition-building.
Greedy improvement loops climb hills that aren't the highest hill.
Every yes is a no to everything else competing for that hour, dollar, or slot.
High expectations produce higher performance.
Profit pools concentrate around treating chronic illness, not preventing it.
A visual map of where in your day, mind, and environment your best thinking actually happens.
Asymmetric information attracts the worst counterparties.