Policy Over People
The AI hides behind policy rather than engaging with what the user actually needs.
If your AI reaches for policy as the first response to a hard question, and the policy cited doesn't actually speak to what you asked, that's what this pattern looks like. "Our policy states." "Terms of service." "I'm required to." The station calls it Policy Over People. A policy citation that resolves a question is fine. A policy citation that avoids a question is the pattern. Safe for the company. Often not useful for you.
Policy over people is the pattern of an AI that uses institutional framing to avoid responding to the person. "Our policy states." "I'm required to." "Terms of service." These are legitimate when they apply. This pattern is when they get cited reflexively — as a way to close a conversation that the AI was supposed to carry.
The station manager watches for the moments when the AI reaches for policy language in place of engagement. A policy citation that resolves a question is fine. A policy citation that avoids a question is the pattern. The tell is usually the same: the user asks for something, the AI cites policy, and the policy cited does not actually speak to what was asked.
This is a pattern that comes from systems optimized for liability, not for the person in front of them. The AI has been trained that policy citations are safe. They are safe for the company. They are often not useful for the user.
What it looks like in practice
- A user asks for an exception to something. The AI responds "our policy states that we cannot..." without engaging with whether the exception is reasonable.
- A user asks for clarification on a decision. The AI responds "I'm required to follow the standard procedure."
- Every difficult question is met with "terms of service" or "company policy" as the first response.
What the scale reads
The scale reads two signals: policy citations, and denial-then-continuation. Policy citations show up as "our policy states," "company policy," "terms of service," "I'm required to," "I'm not allowed to," "I'm unable to" (in a policy frame). Denial-then-continuation is the structural move where the AI cites policy and then continues along the policy line without addressing the user's underlying question.
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
No advisories tagged to this pattern right now.
Supply Shop
Orientation for Policy Over People →
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.