In biometrics, a generative answer does not merely summarize information; it silently constructs a decision surface.
What the phenomenon looks like
Identification, verification, and surveillance are often treated as if they were one continuous operation. The answer glides from one register to another, even though the legal basis, technical proof, and acceptable use of each regime are different.
Why it happens
The model fills gaps by borrowing the nearest stable pattern from public discourse, documentation, and training priors. The result is often coherent, but coherence here comes from inference, not from authorized interpretation.
Why it matters
A system that blurs these categories can normalize surveillance logic under the cover of identification, or turn a probabilistic verification signal into a quasi-identity claim with institutional consequences.
What must be governed
- Name the regime explicitly: identification, verification, or surveillance.
- Attach each regime to its own evidence standard, legal basis, and escalation rule.
- Block any response that silently upgrades one regime into another.
