Guest post: Frame by Frame – a follow-up on AI, rights and risk in animation

Overview

In September, I presented “Frame by Frame: AI updates, rights, risks and best practice”. This article follows on from that session. It summarises where things stand on AI regulation and risk, and sets out practical steps that studios, production companies and freelancers can take now.

AI is already part of animation work. For most organisations, the question is how to use it in a way that protects projects, clients, people and intellectual property. That is where governance comes in. It does not need to be complicated. It needs to be clear, and it needs to be in place. 

Context

How AI is being used

AI covers a wide range of tools. At one end are everyday uses such as transcription, meeting notes and drafting emails. At the other are AI features built into production software, studio-trained models and agents that take actions on your systems. The further along that range you go, the more access the tool needs and the more oversight it requires.

AI tools are being used at several stages of animation work, including pitch and development, storyboarding and previsualisation, and dialogue and localisation. In production, they are being used to assist with clean-up and in-betweening, rotoscoping, upscaling and denoising. They are also used for day-to-day administration.

Some areas carry more risk. Final creative output, such as character design and key art, carries the most IP and credit risk. Scripts, unreleased assets and budgets should never go into free tools. Any AI voice work must start with performer consent and contract terms. Crew agreements and union positions on AI are still developing.

Transparency obligations under the EU AI Act

The transparency obligations in Article 50 of the EU AI Act have applied since 2 August 2026. Fines can reach €15 million or 3% of worldwide annual turnover. For SMEs, including start-ups, the cap is whichever is lower.

Two duties are particularly relevant to animation. Providers of AI systems that generate audio, image, video or text must ensure outputs are marked in a machine-readable format. Generative AI systems on the market before August 2026 have until 2 December 2026 to meet this duty. Deployers, meaning organisations using AI systems in a professional capacity, must disclose when image, audio or video content constituting a deepfake has been artificially generated or manipulated.

For artistic, creative, satirical or fictional work, the obligation is lighter but it does not disappear. Disclosure must be made in a way that does not hamper the display or enjoyment of the work.

In July 2026, the European Commission adopted non-binding Guidelines on Article 50, alongside a voluntary Code of Practice on the transparency of AI-generated content. The Commission has also published EU icons for labelling AI-generated content. These icons are not mandatory, but widespread use is expected.

The Irish picture

The AI Office of Ireland was established in July 2026 under the Regulation of Artificial Intelligence Act 2026. It coordinates Ireland’s approach, while 15 existing regulators supervise AI within their own sectors. Coimisiún na Meán covers media and online platforms, and the Data Protection Commission covers personal data.

Copyright and IP

Courts in the US and Europe are examining both how AI models are trained and what they produce. The case closest to animation is the action brought by Disney, Universal and DreamWorks against Midjourney in June 2025. Warner Bros. brought a separate action in September 2025, and the two were consolidated. The case challenges both how the tool was trained and the images of the studios’ characters it generates. At the time of writing, there has been no ruling on infringement or fair use, and no trial date has been set. Several other cases in the US, UK and Germany are under appeal. The position is evolving. 

The risks

AI output can be fluent, polished and wrong. The tool does not flag when it is unsure, so every output needs review. AI can also produce more material than a team can check, which creates backlogs rather than savings. When a poor output goes out under your name, accountability stays with you, not the tool.

The larger risks for this industry concern data and IP. Free versions of tools such as ChatGPT, Copilot and Gemini can, by default, use your inputs and outputs to train the model, and provider staff may review that data. Client artwork or scripts pasted into these tools leave your control.

The Data Protection Commission’s Case Studies 2025 includes an example where an employee uploaded 32 CVs to a free AI tool from a work computer. The organisation reported it as a personal data breach. It had no policy on free or external AI tools at the time, and the employee was trying to do the job faster. It shows how easily free tools are used without an organisation’s knowledge.

Freelancers should also look closely at the contracts they sign. Check what the AI clauses say about your work. Check the renewal terms, as what you signed a year ago may not be what is in force today. Be clear on what rights you keep.

Practical takeaways: what to do next

  1. Find out what is already in use. Start with an AI inventory: which tools, used by whom, and for what. Without a clear approach, staff pick their own tools, and the organisation has no visibility of it.
  2. Agree your approach. A short AI strategy sets out your principles and priorities, and names who owns AI decisions. Brown Bag Films’ guidelines, for example, include an internal AI Council to approve and monitor AI use.
  3. Write and share an AI policy. Staff need to know what is permitted, for which use cases, what is prohibited and when to check. Review your policy regularly and make sure people understand it.
  4. Keep client and project material in approved tools. Free tools should only be used in limited contexts. For project, client or IP-sensitive material, use enterprise tools with contractual safeguards and no training on your inputs. I
  5. Plan for disclosure. Decide who signs off disclosure wording and where it appears. Keep a record of which projects used generative AI.
  6. Review contracts. Check supplier terms for data protection and IP clauses, and check client, crew and freelancer agreements for AI terms.
  7. Train your people. AI literacy training should be role-specific and refreshed on a regular basis.

Three questions from the session are worth asking yourself and your team:

  1. If a client asked whether you use AI on their project, how would you answer?
  2. If an AI tool gave you a confidently wrong answer, would you catch it before it reached a client? 
  3. Who in your organisation owns AI decisions, and how will you review them over time?

Elaine Morrissey
Rock Consultancy, Privacy & AI Governance
info@rockconsultancy.ie | www.rockconsultancy.ie

Further reading