| name | genai-disclaimer |
| description | Standard disclaimer and attribution templates for AI-generated or AI-assisted data analysis outputs |
GenAI Disclaimer
Standard disclaimers and attribution notices for AI-generated or AI-assisted data analysis outputs.
When to Apply
Add a GenAI disclaimer to any output that:
- Was generated or substantially shaped by an AI/LLM
- Contains AI-interpreted statistics, summaries, or conclusions
- Includes AI-generated Data Dictionaries, Descriptions, or Tags (from
describegpt)
- Was validated, profiled, or cleaned with AI-assisted workflow decisions
- Contains visualizations where chart type, aggregation, or framing was AI-selected
Standard Disclaimer Templates
Short (inline footnote)
*Analysis assisted by AI. Results should be independently verified before decision-making.*
Medium (report footer)
---
**AI Disclosure:** This analysis was generated with AI assistance using qsv data wrangling
tools and Claude. All statistical computations were performed by qsv (deterministic,
reproducible); interpretations, summaries, and recommendations were AI-generated and should
be reviewed by a domain expert before acting on them.
Full (formal deliverable)
---
## AI-Generated Content Disclosure
**Tools used:** qsv (data processing), Claude (analysis and interpretation)
**Date:** [YYYY-MM-DD]
This document contains AI-generated content. The following aspects were AI-assisted:
- [ ] Data profiling interpretation and quality assessment
- [ ] Statistical summary and narrative
- [ ] Data Dictionary, Description, and Tags
- [ ] Chart type selection and visualization design
- [ ] Recommendations and conclusions
**What is deterministic:** All row counts, statistics, frequency distributions, and data
transformations were computed by qsv — a deterministic, open-source tool. Given the same
input data and parameters, these results are exactly reproducible.
**What is AI-generated:** Narrative interpretations, quality assessments, chart design
decisions, and recommendations were generated by an LLM. These outputs may contain errors,
hallucinations, or biases inherent to the model. They should be reviewed by a qualified
analyst before use in decision-making.
**Reproducibility:** The qsv commands and parameters used are documented in this analysis.
Re-running them on the same input data will produce identical numerical results. AI-generated
narratives may vary between runs.
Choosing the Right Level
| Output Type | Recommended Level |
|---|
| Quick Slack message or internal chat | Short |
| Internal report or Data Schematic annotation | Medium |
| Stakeholder presentation | Medium or Full |
| Regulatory or compliance deliverable | Full |
| Published research or external report | Full |
| Exploratory analysis (own use) | None needed |
What to Attribute to AI vs. qsv
| Component | Attribution | Reproducible? |
|---|
| Row counts, statistics, frequencies | qsv (deterministic) | Yes — exact same results every run |
| Data type inference, null counts | qsv (deterministic) | Yes |
| Data cleaning operations (dedup, trim, safenames) | qsv (deterministic) | Yes |
| SQL query results via sqlp | qsv/Polars (deterministic) | Yes |
| Data Dictionary, Description, Tags | AI-generated via describegpt | No — may vary between runs |
| Quality assessment narrative | AI-generated | No |
| Chart type recommendations | AI-generated | No |
| Conclusions and recommendations | AI-generated | No |
| Validation methodology review | AI-generated | No |
Integration with Commands
When these commands produce output, consider appending the appropriate disclaimer:
/data-profile → Medium or Full (contains AI interpretation of statistics)/data-describe → Full (describegpt output is entirely AI-generated)/data-validate → Medium or Full (methodology review is AI-generated)/data-viz → Medium (chart type selection and title framing are AI decisions)/data-clean → Short or None (operations are deterministic; only the choice of operations was AI-guided)/data-join → Short or None (join execution is deterministic)/csv-query → Short or None (SQL results are deterministic)/data-convert → None (purely mechanical format conversion)
Regulatory Considerations
Some jurisdictions and industries require explicit AI disclosure:
- EU AI Act: Transparency obligations for AI-generated content
- Financial services: Regulatory expectations around AI-assisted analysis and model risk
- Healthcare: HIPAA considerations when AI processes PHI; disclosure requirements for AI-assisted clinical decisions
- Government: Executive orders and agency policies on AI use in federal data analysis
When in doubt, use the Full disclaimer. Over-disclosure is always safer than under-disclosure.
