Explain how copyright law applies differently to generative AI training data and to a model's generated output.
Digital Transformation & Artificial Intelligence
Copyright and Licensing Risks in Generative AI Datasets and Outputs
Learn to assess copyright and licensing risk across generative AI training data, model outputs and vendor contracts, and build practices that protect against infringement claims.
🗓️Duration5 training days
📚Content4 modules · 8 sessions
🎓On completionAccredited attendance certificate
About the programme
Course Overview
Generative AI models are trained on vast quantities of text, images and code whose copyright status is often unclear, and the outputs those models produce raise a second, separate question about ownership and originality that many organisations have not yet resolved. Publishers, rights holders and regulators are actively testing these questions through litigation and guidance, which means the risk position for any organisation using generative AI in commercial work keeps shifting. This course gives participants a working framework for assessing copyright exposure across the generative AI lifecycle: how text and data mining exceptions apply to training data, how to build a dataset provenance record, how to evaluate a vendor's licensing and indemnification terms, and how to judge whether a specific output is original enough to attract copyright protection. Participants also learn to screen outputs for similarity to existing works and signs of training-data memorisation before publication. The course closes with the policy and incident-response practices that let an organisation use generative AI commercially while limiting its exposure to infringement claims and reacting quickly when a dispute arises.
Expected Learning Outcomes
Assess whether a dataset collection method falls within a text and data mining exception or requires a licence.
Build a dataset provenance record that documents source, licence terms and collection method for audit purposes.
Negotiate vendor indemnification and licensing terms that allocate liability for third-party infringement claims.
Judge whether an AI-generated output qualifies for copyright protection based on the human authorship involved.
Screen outputs for similarity and memorisation risk before publication using recognised detection methods.
Respond to a copyright claim or takedown notice with an evidentiary record and a defined escalation path.
Who Should Attend
In-house counsel and IP specialists advising on generative AI adoption and licensing.
Content, marketing and design leads deploying generative AI tools in commercial output.
Procurement teams negotiating contracts with generative AI and model providers.
Data scientists and machine learning engineers responsible for training data selection.
Product managers building features that generate text, images or code for customers.
Compliance officers building acceptable-use policy for generative AI across the organisation.
Course Modules
Select any module to see its sessions and points.
01Copyright Foundations for Generative AI Systems
2 sessions · 8 points
Session 1How Copyright Law Applies to Training Data and Outputs
- Explain how copyright protects original expression and why unauthorised reproduction during model training can infringe that protection.
- Distinguish the legal treatment of ingesting copyrighted works for training from the legal treatment of a model's generated output.
- Compare how different jurisdictions currently treat the use of copyrighted material to train generative AI models.
- Assess the copyright status of outputs generated with minimal human input against those substantially shaped by human creative choices.
Session 2Text and Data Mining Exceptions and Their Limits
- Explain the scope of text and data mining exceptions available in relevant jurisdictions and the conditions attached to them.
- Identify how rights holders can reserve their rights and opt out of text and data mining under applicable exceptions.
- Assess whether a specific training pipeline's data collection method falls within a lawful exception or requires a licence.
- Track pending litigation and regulatory guidance that could narrow or extend current text and data mining exceptions.
02Assessing Risk in Datasets and Model Providers
2 sessions · 8 points
Session 1Dataset Provenance and Due Diligence
- Build a dataset provenance record that documents the source, licence terms and collection method for each data source used.
- Screen training datasets for copyrighted, trademarked or otherwise restricted material before a model is fine-tuned on them.
- Assess the risk profile of web-scraped datasets against curated, licensed or synthetic alternatives for a specific use case.
- Document data lineage so an organisation can respond to a rights holder's takedown or provenance enquiry after deployment.
Session 2Evaluating Vendor Contracts and Indemnification
- Review a generative AI vendor's terms of service to identify what training data warranties and indemnities are actually offered.
- Negotiate indemnification clauses that allocate liability for third-party infringement claims arising from model outputs.
- Assess licence terms that govern reuse, redistribution and commercialisation of outputs generated through a third-party tool.
- Compare open-weight model licences against proprietary API terms for restrictions on commercial and derivative use.
03Managing Output Risk and Ownership
2 sessions · 8 points
Session 1Ownership and Originality of AI-Generated Outputs
- Apply current guidance on the copyrightability of AI-generated works to determine what protection, if any, an output receives.
- Assess how much human authorship, editing or selection is needed for an output to qualify for copyright protection.
- Draft internal guidance on attributing and registering works that combine human and AI-generated contributions.
- Evaluate contractual assignment clauses that determine who owns outputs produced by employees or contractors using generative tools.
Session 2Similarity, Memorisation and Infringement Screening
- Use similarity-detection tools to screen generated text, images or code against known copyrighted works before publication.
- Recognise signs of training-data memorisation, where a model reproduces near-identical passages, images or code from its training set.
- Assess substantial-similarity risk for outputs that closely track a specific existing work in style, structure or expression.
- Build a pre-publication review gate for high-risk outputs, such as commercial creative work or branded content.
04Governance, Policy and Dispute Response
2 sessions · 8 points
Session 1Building Organisational Policy and Controls
- Draft an acceptable-use policy that tells staff which generative AI tools are approved and what disclosure is required for outputs.
- Define retention and audit trails that record the prompts, model version and sources used to produce a published output.
- Train content, design and engineering teams to recognise and escalate outputs with elevated copyright risk.
- Align generative AI policy with existing brand, trademark and licensing clearance processes already used by the organisation.
Session 2Responding to Claims and Disputes
- Build an escalation path for a copyright claim or takedown notice affecting AI-assisted content already published.
- Coordinate with legal counsel to assess exposure when a vendor's indemnity is limited, excluded or disputed.
- Prepare an evidentiary record, including prompts and generation logs, to support a defence against an infringement claim.
- Review and update licensing and vendor selection decisions based on lessons from a claim or a near-miss incident.
What the participant receives
4 course modules
A structured syllabus
8 training sessions
across 5 days
32 detailed points
Applied, detailed content
Accredited attendance certificate
On completing the programme
Complete your registration
We will contact you within one business day to confirm.
Ready to start?
Reserve your seat and start building the skill.