skills/policy-ai-ip-and-copyright/SKILL.md
Generates a formal agency policy document on intellectual property and copyright for AI-assisted creative work. Covers IP ownership of AI-assisted deliverables, copyright qualification thresholds under UK/US/EU law, disclosure obligations, provenance tracking, and the standard for professional authorship. Invoke when a client asks who owns AI-assisted content, whether AI-generated deliverables can be copyrighted, what the agency's disclosure obligations are, or when onboarding a client who will receive AI-assisted work.
npx skillsauth add peterbamuhigire/social-media-skills policy-ai-ip-and-copyrightInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
SKILL.md; do not skip mandatory steps or required fields.references/ directory is added later, treat its files as the deeper source material and keep this SKILL.md execution-focused.This skill produces a standalone policy document that the agency can hand directly to clients when questions arise about intellectual property ownership, copyright eligibility, and disclosure of AI-assisted work. The document is based on current UK, US, and EU legal frameworks as summarised in Ching and Mothi (2025) AI for Creatives: Unlocking Expressive Digital Potential (CRC Press).
Scope note: This policy summarises the legal landscape as understood at the time of publication. It does not constitute legal advice. For contract-specific IP clauses or disputes, refer the client to a qualified IP solicitor.
Ask for the following before generating the policy document:
When the agency uses AI tools to produce content on behalf of a client, who holds the intellectual property? The answer depends on three factors:
The agency retains copyright in all deliverables produced using AI tools where a human editor, strategist, or writer has made substantive creative decisions in the process — including selection, arrangement, modification, and editorial judgement. Such deliverables are then assigned or licensed to the client under the terms of the engagement contract.
Where a deliverable is generated substantially without human creative input, no copyright protection is available under current law in the UK, US, or EU (see Part 2). In those cases, the deliverable enters the public domain and cannot be protected.
Practical implication: The agency's standard workflow applies human review and editorial modification to all AI-assisted deliverables before client delivery. This is both a quality standard and a legal necessity.
United Kingdom — Copyright, Designs and Patents Act 1988 UK law is unique in recognising "computer-generated works" where there is no human author. Section 9(3) grants copyright to "the person by whom the arrangements necessary for the creation of the work are undertaken." However, the threshold for what qualifies as sufficient "arrangement" is untested in UK courts for generative AI. The safe position is to ensure substantive human creative input at every stage. Works created solely by machine with no human creative selection or arrangement do not qualify for protection under the standard literary/artistic works provisions.
United States — Title 17, §102(a), US Copyright Act The US Copyright Office has issued formal guidance (2023) that copyright protection does not extend to works created solely by AI. Human authorship is a constitutional and statutory requirement. The Office has granted protection to human-selected and arranged components of AI-assisted works, but refused protection for AI-generated elements taken in isolation. The threshold test is whether a human made sufficiently creative choices in the selection, co-ordination, or arrangement of the final work.
European Union — Directive 2019/790 on Copyright in the Digital Single Market EU copyright law requires human authorship for protection. AI-generated content with no human creative contribution is not eligible for copyright under the Directive. The EU AI Act (2024) adds further obligations (see Part 3) but does not create new IP rights for AI outputs.
For the agency's deliverables to qualify for copyright protection in any of the above jurisdictions, the producing human must have made at least one of the following:
A human who merely presses a button and accepts unmodified AI output without creative judgement has not met this threshold.
The Writers Guild of America's 2023 Master Basic Agreement established a professional authorship standard that the agency adopts as its benchmark:
Agency application: The agency's name and the name of any credited human creator may only appear on a deliverable where a human has made substantive creative decisions. AI is used as a production tool, not as a credited author.
Under the EU AI Act, providers and deployers of AI systems must ensure that content generated by AI is labelled as such when it is likely to be perceived by an audience as authentic human-created content. This obligation applies to:
Agency standard — Specific, Not Vague Disclosure
Vague disclosure ("made with AI") does not meet the spirit of the Act or the agency's professional standard. Apply the following disclosure language where relevant:
"This [content type] was produced using AI-assisted tools under human editorial direction. The creative brief, final selection, and editorial decisions were made by [Agency Name]."
Where AI has generated only supporting or draft elements and the final published work is substantially human-revised, apply:
"AI tools were used in the drafting process. All published content has been reviewed, edited, and approved by [Agency Name]."
For high-risk AI applications, Article 28b(4) requires that humans retain meaningful oversight and the ability to override AI decisions. The agency's editorial review process is designed to satisfy this requirement for all content deliverables.
As AI-generated content becomes indistinguishable from human-created content, clients face reputational, legal, and commercial risks if the provenance of their content cannot be verified. Provenance tracking creates an auditable record of how content was produced.
SynthID is Google DeepMind's standard for watermarking AI-generated audio and images at the point of generation. Watermarks are embedded in the content itself and can be detected even after standard post-processing. The agency recommends:
For each AI-assisted deliverable, maintain an internal production record that captures:
Retain production records for a minimum of three years. Make them available to clients on request and to legal counsel in the event of an IP dispute.
Refer the client to a qualified IP solicitor in the following circumstances:
"This document summarises the agency's policy and our understanding of the current legal framework. It is not legal advice. For contractual IP warranties, copyright registration, or dispute resolution, please consult a qualified intellectual property solicitor with experience in AI and digital content law."
A well-produced policy document under this skill:
tools
Generates a foundational social media training guide for clients and their teams who are completely new to social media marketing, or who have been posting without any strategic understanding. Invoke when the user says "write a social media basics guide", "create a beginner training document", "the client doesn't understand social media", "start-here training", or when a client needs to understand social media before any strategy or content work begins. Distinct from training-client-team (operational handover of an existing strategy) and training-diy-content (content creation for self-managing clients). This skill covers what social media is, how it works, and how to approach it intelligently — the conceptual foundation that makes all downstream strategy work land.
tools
Generates a practical smartphone video production training guide for East African clients and content teams. Covers shooting, audio, lighting, framing, editing, and platform-specific formats using only a smartphone — no professional equipment required. Invoke this skill when a client or their team needs to produce their own social video content and requires a hands-on, jargon-free training document tailored to EA field conditions.
tools
Generates a complete DIY content creation handbook for clients who want to manage some or all of their own content after the initial strategy engagement. Invoke when the user says "write a DIY content guide", "create a self-managed content handbook", "the client wants to manage their own content", or when a handover guide is needed at the end of a strategy engagement. Output is a self-contained reference document — not a training presentation — that the client keeps and uses independently.
tools
Generates a complete 2-hour in-person training workbook for a client's internal team — employees who will assist with content creation or community management. Invoke when the user says "create a team training guide", "write a staff training workbook", "onboard our internal team on social media", or needs a printable workshop document for client employees. Output is a structured, print-ready workbook — not a presentation deck.