Direct answer: if your team uses AI as a tool and a person makes the creative choices — choosing the angle, rewriting, selecting and arranging — a hotel room description, tour itinerary or edited photograph is protected like any other original work, and the copyright normally belongs to your business if an employee made it (CDPA 1988 s11). If the AI produced the text or image with no human author, UK law has an unusual extra rule: the author is “the person by whom the arrangements necessary for the creation of the work are undertaken” and protection lasts 50 years (CDPA s9(3); s12(7)). But the government said in March 2026 that this rule is legally uncertain, has little evidence of use, and that it proposes to remove it (Report on Copyright and AI, Section I). So do not build your content strategy on owning raw AI output. Own the parts you can prove are yours.
Kaize is a UK AI product studio for travel and hospitality. This article is based on the Copyright, Designs and Patents Act 1988 (CDPA), the government’s Report on Copyright and Artificial Intelligence and its Impact Assessment, both published on 18 March 2026 and presented to Parliament under the Data (Use and Access) Act 2025 (s136), the High Court judgment in Getty Images v Stability AI and Taylor Wessing’s note on the appeal, and the published terms or commitments of Anthropic, Canva, Microsoft and Google Cloud, all read on 21 October 2026. It covers what you can own and what you might infringe when you publish AI-assisted marketing content. It is not legal advice, and it does not describe any Kaize client project.
Why this is a travel and hospitality question
Travel businesses publish more descriptive content per pound of revenue than most sectors: room and suite descriptions, villa listings, day-by-day itineraries, destination guides, menus, brochures, email campaigns and the photographs that sell all of them. That is exactly the content generative AI is now used to draft, translate and retouch. Three practical questions follow. If a competitor or an aggregator copies your AI-assisted itinerary word for word, can you stop them? If your agency or freelancer delivers AI-generated work, who owns it? And if the AI reproduces someone else’s photograph, text or logo, who is liable when you publish it?
Three kinds of AI output, three different answers
1. AI-assisted work: normal copyright
Copyright protects original literary and artistic works, including photographs (CDPA s3; s4). The government’s report says that any work “made using AI as a tool, but where the creative expression in the work comes from a human creator” will generally meet the originality requirement — for example “an original photograph which is taken by a human photographer and edited using an AI tool, or an original literary work which is edited with help from an AI assistant” (Report, Section I). It calls these “AI-assisted works” and notes they are protected on similar terms in the EU and USA. You do not register copyright in the UK; it arises automatically (GOV.UK: How copyright protects your work).
Who owns it depends on who made it. The author is the first owner, but where an employee makes a work in the course of employment, the employer is the first owner unless agreed otherwise (CDPA s11). That rule does not cover agencies or freelancers: copyright stays with them unless it is assigned, and an assignment is only effective if it is “in writing signed by or on behalf of the assignor” (CDPA s90(3)).
2. Wholly AI-generated work: the uncertain UK rule
The CDPA defines “computer-generated” as generated by computer “in circumstances such that there is no human author of the work” (s178). For those works the author is the person who made the necessary arrangements (s9(3)) and copyright lasts 50 years from the end of the year the work was made (s12(7)). The report says that for a general-purpose AI responding to a prompt, the author “will usually be the person who inputted the prompt”. But it also sets out the problem: s9(3) only protects works that are original, and originality is tested as an “author’s own intellectual creation” with a “personal touch” — which suggests a work with no human author could struggle to qualify. The report says it is “unclear in the absence of case law how an ‘original’ yet wholly machine-authored work would be defined” (Report, Section I).
Most other jurisdictions give no specific protection here, including the USA and the EU, according to the same report. The government’s conclusion: there is “minimal evidence” the protection is used, 78% of online survey respondents who answered did not want it kept, and “in the absence of evidence of its ongoing value, we propose that it should be removed” (Report, Section I). That is a proposal, not a change in the law. Nothing has been removed yet, and the report also says the government “will not introduce reforms to copyright law until we are confident that they will meet our objectives” (Report, Section A).
3. Output that copies someone else’s work: possible infringement
Copying the whole or a substantial part of a protected work without permission is infringement (CDPA s16; s17). The report is direct about AI outputs: if a service generates a substantial part of a copyright work, “both the user and provider of a model may be liable”, for example “if a user specifically requests an AI service to reproduce an artist’s work and the service does so”. It also says that if you feed someone else’s work into an AI tool, such as a photo editor, “they will need to ensure they have the relevant permissions to use it” (Report, Section B).
What the courts have said so far
The only UK judgment on AI training so far is Getty Images v Stability AI, decided by the High Court on 4 November 2025 ([2025] EWHC 2863 (Ch), judgment). The government summarises it this way: a trained model can in principle be an “article” and so an infringing copy, but because there was no evidence that Stable Diffusion contained copies of Getty’s works, it was not (Report, Section B). Getty has permission to appeal that point; Stability was refused permission to appeal the limited and historic findings of trade mark infringement, and in January 2026 Taylor Wessing expected the appeal might be heard at the end of 2026, with a decision perhaps a year after that note (Taylor Wessing, January 2026). The case is about training, not about whether a hotel owns its AI-written copy. For operators the practical lesson is narrower: outputs that show another brand’s marks or watermarks are a risk in themselves.
What the AI vendors’ terms say
Ownership between you and the vendor is a contract question, separate from whether copyright exists at all. Read the terms for the tool and the plan you actually use; consumer and business terms often differ.
- Anthropic’s Commercial Terms say the customer “owns its Outputs” and Anthropic assigns to the customer “its right, title and interest (if any)” in them; for paid use it will defend customers against third-party claims that outputs infringe intellectual property rights (Anthropic Commercial Terms). Note the words “if any”: a vendor can only assign what exists.
- Canva’s AI Product Terms say you own your Output, except Output that modifies or incorporates Canva’s licensed content, and warn that “outputs may not be unique, and other users may receive similar Outputs” (Canva AI Product Terms). Two hotels can get near-identical copy from the same prompt.
- Microsoft’s Customer Copyright Commitment says it will defend commercial customers and pay adverse judgments if they are sued for copyright infringement over Copilot and Azure OpenAI Service outputs, but only if they have implemented the required guardrails and mitigations (Microsoft Customer Copyright Commitment; required mitigations).
- Google Cloud offers a training-data indemnity and a separate generated-output indemnity for listed Workspace and Cloud services (Google Cloud). Check whether the product you use is on the covered list.
Indemnities are useful, but they protect against claims that you infringed someone else. They do not give you a right to stop a competitor copying your content.
Eight scenarios for hotels, tour operators and DMCs
Questions to ask your AI vendor or agency
- Under the plan we are on, who owns inputs and outputs, and is there any licence back to you?
- Do you indemnify us against third-party IP claims over outputs, for which products, and what mitigations must we have switched on?
- Do you train on our inputs or outputs, and can we opt out?
- Can the tool block or flag outputs that reproduce trade marks, watermarks or long passages of existing text?
- Agency: which deliverables used generative AI, which tools, and will you assign copyright in writing and warrant that the work does not infringe?
- Can we export prompts, drafts and edit history so we can show human creative input later?
A 9-step content control for AI-assisted marketing
FAQ
Does my business own text that ChatGPT, Claude or Gemini writes for us?
Contractually, most business terms give you the vendor’s rights in the output. Whether copyright exists to own depends on human creative input: AI-assisted work is protected as normal; wholly AI-generated work relies on the uncertain s9(3) rule (Report, Section I).
Is the UK about to change the law?
The government proposes removing protection for computer-generated works but has not legislated, and says it will not reform copyright until it is confident reforms meet its objectives (Report on Copyright and AI; Impact Assessment).
Do we have to label AI-generated content in the UK?
The report says there are no UK obligations to label AI-generated content and proposes working with industry on best practice (Report on Copyright and AI, summary of proposals). Advertising rules on misleading content still apply, and EU rules may apply to EU-facing services — see our articles on AI-edited photos and EU AI Act Article 50.
Can we use AI to rewrite our old brochures?
Yes, if the business owns the brochure copy or has a licence that covers it. If an agency or freelancer wrote it, check what was assigned in writing before feeding it into an AI tool.
Who is liable if AI output infringes someone else’s copyright?
Possibly both the user and the provider, depending on the circumstances (Report, Section B). A vendor indemnity may cover you, but only on the terms and products it names.
Kaize point of view
For travel businesses, the defensible asset was never the adjectives. It is the knowledge behind them: which room actually has the sea view, which ferry connects with which tour, what the guide does when it rains. AI makes generic travel prose almost free, and the law is moving away from protecting prose with no human author. So use AI to draft and translate, but put your verified facts, your own photography and your people’s judgement into every piece you publish, and keep the record that shows it. That is what makes the content worth copying, and what gives you a basis to object when someone does.
Related reading: AI-edited and generated hotel and villa photos and the ASA, AI-translated hotel and tour operator websites, what staff can put into ChatGPT, Copilot or Gemini, controlling AI crawlers with robots.txt, EU AI Act Article 50 disclosure and whether to build or buy travel software.
If you run a hotel group, tour operator or DMC and are scaling AI-assisted content, Kaize can help you set up the source data, review steps and records that keep it accurate and yours. Start with an AI opportunity review or discuss a project with Kaize.