Operational evidence, not disclosure theatre.
This sample maps five invented agency AI use cases to role-based literacy measures, a conservative Article 50 screen and objective implementation work. It deliberately does not label every AI-assisted output.
- Use cases
- 5
- Role groups
- 3
- Screens
- 5
- Tasks
- 7
Use-case screen
Initial operational classifications require customer verification and professional review where marked.
| Use case | Exposure | Initial screen and rationale | Next action |
|---|---|---|---|
| UC-01 Advertising copy assistant | Client and public after approval | No action identified Ordinary advertising draft with mandatory human editorial control is not automatically a disclosure case; literacy and client-approval controls still apply. | Retain editor approval evidence |
| UC-02 Website support chatbot | Direct visitor interaction | Action A person directly interacts with an AI system. Verify whether the provider's product supplies a clear notice at first interaction and who contractually owns it. | Add/verify first-interaction notice and human escalation |
| UC-03 Synthetic presenter video | Public campaign audience | Action A realistic synthetic presenter can constitute deep-fake content; disclose artificial generation in an appropriate and accessible manner without hampering the work. | Place disclosure with video and record client approval |
| UC-04 Translation assistant | Client after human approval | No action identified Routine translated copy is human reviewed and is not public-interest text in this fictional case. No blanket AI label is recommended. | Retain translation review evidence |
| UC-05 Public-interest trend article | Public readership | No action identified The fictional process includes human review/editorial control and a named legal person holding editorial responsibility. Confirm those facts before relying on the exception. | Retain source and editorial sign-off; obtain legal review before publication |
Prioritized implementation backlog
| Priority | Task | Owner | Acceptance criteria |
|---|---|---|---|
| Critical | Verify chatbot first-interaction notice and provider/deployer responsibility | Digital Lead | A clear accessible notice is visible by first interaction; contract/product evidence names the responsible party |
| Critical | Add synthetic-presenter disclosure | Creative Director | Disclosure appears with the video and is readable/audible as appropriate without obstructing the work |
| High | Record mandatory editor approval for published ad copy | Creative Director | Every published variant identifies editor and approval date |
| High | Document native-level translation review | Account Director | Each delivered translation has reviewer and approval date |
| Critical | Confirm public-interest article editorial-control exception before publication | Strategy Director | Named editor and legal entity accept editorial responsibility; professional reviewer records conclusion |
| High | Run role-based literacy initiative | Managing Director | All three role groups receive module; attendance/initiative record exists; check target is 80% |
| Medium | Schedule quarterly review | Managing Director | Calendar owner and 2026-10-01 review event recorded |
Included in the pilot
- Use register and role matrix
- Tailored literacy module
- 12-question knowledge check
- Initiative records
- Five external-use screens
- Disclosure options and task backlog
Not represented
- Legal advice or certification
- Definitive provider/deployer classification
- High-risk-system assessment
- Technical provenance implementation
- Customer results or testimonials