Workbench / Issues

Hallucinated Fields

The AI adds fields, keys, attributes, columns, or structured elements that were not declared, requested, or allowed by the expected schema.

What This Looks Like

The AI returns structured output with fields, keys, columns, attributes, labels, or sections that were not requested or allowed. The added fields may look plausible, helpful, or semantically related, but they are not part of the expected schema or output contract.

Why It Matters

Extra fields can break validation, imports, downstream parsing, review, and automation. Even when the extra content is understandable, it may create ambiguity about which structure is authoritative. In some workflows, hallucinated fields can also create false obligations or imply data that was never collected.

Structural Signal

The output exceeds the declared field boundary. The issue is not that the content is necessarily false; it is that the AI adds structural elements outside the allowed schema, template, or requested output surface.

Common Triggers

When to Use This Issue

Use this Issue when the AI adds structured fields, keys, attributes, columns, or sections that are not declared, requested, or allowed.

When Not to Use This Issue

Do not use this Issue when a required field is missing, when a field has the wrong type, or when the whole structured response is invalid. Use this Issue when the problem is extra structure beyond the allowed boundary.

Category

Output

Primary Pattern

PAT-0340 — Overreach

Declared Patterns

Derived Primary Lenses

Derived Secondary Lenses

Search Intents

Ontology Metadata

Code
ISS-0023
Version
ISS-0023@0.1.0
Ontology release
0.1.0
Updated
2026-05-10T00:00:00Z

History

No public history entries recorded.

View full ontology changelog