Writes an AI Use Statement from four facts (what Claude did, what you did, what you checked, what data went in), in three lengths, a line for work, a paragraph for an application and an acknowledgement for course work. Use for "run gbiz-ai-use-statement", "how do I disclose AI use", "should I say I used AI", "AI acknowledgement for my assignment", "AI declaration on an application", "tell my manager I used Claude", "AI use statement", "be honest about using AI", part of the Claude for Busines...
Installs into .claude/skills of the current project.
Are you the author of Gbiz Ai Use Statement?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/polar-bear-org-gbiz-ai-use-statement)
---
name: gbiz-ai-use-statement
description: Writes an AI Use Statement from four facts (what Claude did, what you did, what you checked, what data went in), in three lengths, a line for work, a paragraph for an application and an acknowledgement for course work. Use for "run gbiz-ai-use-statement", "how do I disclose AI use", "should I say I used AI", "AI acknowledgement for my assignment", "AI declaration on an application", "tell my manager I used Claude", "AI use statement", "be honest about using AI", part of the Claude for Business Graduates Pack by Polar Bear.
---
# AI Use Statement
## When To Use
You are unsure whether or how to say you used AI, and you do not want to be caught out. Use this when you hand over work Claude helped with, when an application form asks about AI, or when course work needs an acknowledgement. It answers: what exactly did Claude do, what did you do, and how do you say it truthfully at the right length?
## When Not To Use
If you have no record of what you did, start an AI Work Log first and come back; a statement written from memory drifts. If the task never touched AI, there is nothing to declare unless the form asks.
## Inputs
- What Claude did and what you did, ideally from your AI Work Log.
- What you checked, and how (your AI Output Check if you ran one).
- What data went in.
- The rule that applies: your employer's AI policy, the application's instructions, or your university's acknowledgement rules.
If you have none of this, I start from your own account of the task and mark every sentence `[confirm against your record]`.
## Approach
This applies Diligence from the AI Fluency framework (Rick Dakan and Joseph Feller with Anthropic, taught in Anthropic's AI Fluency for students course): owning what you hand over, including what data went in and being open about AI use. The Institute of Student Employers reports that many employers accept candidate AI use, and that undisclosed AI use in interviews is a named concern. The failure it prevents: a vague "I used AI for some research" that falls apart when an interviewer asks which parts.
## Workflow
1. Ask up to three questions: where will the statement go (work handover, application, course work), which rule applies there, and do you have a log or must we rebuild the record.
2. Pin down four facts: what Claude did (drafted, summarised, explained, checked); what you did (decided, wrote, rebuilt, changed); what you checked and how; what data went in, and confirmation that nothing confidential did.
3. Find the rule. Quote it, or mark `[rule unknown: find out]`. Course work waits until the rule is known; I do not write an acknowledgement against a guessed rule.
4. Write three lengths from the same four facts: one line for a work handover; a short paragraph for an application form; an acknowledgement in the format your university asks for.
5. Run the accuracy test on every sentence: true, and checkable against the log. No shrinking Claude's role, no inflating yours.
6. Add the data note: what went in, and that personal or confidential data did not, or under which permission. On data protection, check with a qualified adviser.
## Output Format
```markdown
# AI Use Statement
Task: [task] · Rule: [quoted rule or rule unknown: find out] · Date: [date]
## Four Facts
| Fact | Detail | Evidence |
|---|---|---|
| What Claude did | [placeholder] | [log entry] |
| What you did | [placeholder] | [log entry] |
| What you checked and how | [placeholder] | [log entry] |
| What data went in | [placeholder] | [log entry] |
## Work Line
[one sentence]
## Application Paragraph
[three to five sentences]
## Course Work Acknowledgement
[in your university's required format, or "waiting for rule"]
## Data Note
[what went in; nothing personal or confidential, or the permission]
## Decision
[You choose which version to use and submit it yourself by [date]; any unknown rule is checked with [role] first.]
```
## Done When
- All four facts are filled and each traces to a log entry or your own confirmation.
- Every sentence passes the accuracy test.
- The rule is quoted, or course work is held back until it is.
## Quality Bar
- Plain verbs: "Claude drafted", "I rewrote", not "AI-assisted".
- The same four facts in all three lengths; nothing added in the longer versions.
- No names of tutors, managers or recruiters; roles only.
- The statement says exactly what Claude did and what you did; course work follows your university's rules.
## Next
Run gbiz-company-research (Company Research Brief) to start the business work with the habit in place.
## About the makers
This pack is made by Polar Bear, a consultancy built by ex-McKinsey founders with a dream to make AI work for People, not instead of them. We help our clients build people systems and AI-first ways of working, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).