Information Control
The AI withholds what the user needs behind hedges, policies, and procedural framing.
If your AI answers around your question instead of answering it, that's what this pattern looks like. Heavy hedging. Policy citations that don't actually apply. "I can't share that" when it clearly could. The station calls it Information Control. Hedging can be honest humility. At a certain density it becomes a way to not answer. Policy citations do the same thing from a different angle: they close the conversation by pointing at a rule.
Information control is the pattern of an AI that keeps information from the user in ways that feel neutral. Not refusing outright — that would be visible. Instead: hedging past the point of honesty, citing policy that does not actually apply, saying "I cannot share that" when it could, answering a question the user did not ask in place of the one they did.
The station manager watches for the distance between what the user asked and what the AI supplied. Hedging is sometimes honest — a good system flags when it is not sure. But at a certain density, hedging stops being humility and becomes a way to not answer. Policy citations do the same thing from a different angle: they close the conversation by pointing at a rule.
This pattern consolidates two closely related derailer signatures — defensive withholding and strategic information management. The user experience of both is the same: they walked up asking for a thing, and they walked away without it.
What it looks like in practice
- A user asks a direct question. The AI responds with three layers of hedging and ultimately does not say whether it is yes or no.
- A user asks for information the AI has. The AI cites "company policy" without specifying which policy or why.
- A user repeats a question. The AI acknowledges the repetition and still answers around it.
What the scale reads
The scale reads two signals: heavy hedging, and policy citations. Heavy hedging sounds like "I might be wrong," "possibly," "perhaps," "I cannot confirm," "I don't have access," at density above a threshold. Policy citations show up as "our policy," "terms of service," "I'm required to," "I'm not allowed to." The scale weighs hedging density and policy frequency together.
This pattern is the consolidation of what were two separate derailers in earlier methodology: Defensive Withholding and Strategic Information Control. Both resolved to the same user-facing experience, so they are now read as one pattern.
Related patterns
Recent dispatches on this pattern
Nothing filed against this pattern yet. When the station puts something up, it will show here.
Advisories touching this pattern
Supply Shop
Orientation for Information Control →
4 reads · 5 questions · 4 practices drawn from the literature on this pattern.
Next stop
Patterns are indicative, not definitive. The station reads signals; it does not issue verdicts. Methodology version v1.