Intellectual property, ideas and innovations in a digital world

Sheena Sood, Senior Partner of international law firm Beale & Co, shares her thoughts on artificial intelligence (AI) and Intellectual Property ("IP") as technology continues to advance and policymakers grapple with the complexities of regulating a digital age.

These are areas of particular interest to businesses operating in the design and construction sectors, where technology and innovation, alongside healthy competition and ethical practices, are critical to creating and sustaining the built environment.

With AI available as a resource in many aspects of our working and personal lives, clarity over its use is increasingly important.

The EU’s AI Act has made significant progress since its phased implementation began in 2024. Creating rules regarding the supply and use of AI systems in EU member states, it includes provisions for operators outside of the EU, therefore relevant to companies in the UK and elsewhere working in the EU. The international standard, ISO/IEC 42001:2023, also provides a framework for “establishing, implementing, maintaining, and continually improving an Artificial Intelligence Management System (AIMS) within organisations”. The Alan Turing Institute, in collaboration with Innovate UK BridgeAI, Digital Catapult, STFC Hartree Centre and BSI, also introduced an “actionable” AI Use Case Framework to help “support your AI adoption journey”.

Much to digest, and efforts to date have aimed to provide rules, guidance and support for the use of AI. But, with more to follow, what of protections against its misuse?

Given that 92% of companies plan to increase AI investment in the next three years (McKinsey & Company, Superagency in the Workplace) the race is on to create far-reaching legislative frameworks, internationally and in local jurisdictions, which transcend commercial, creative and safety parameters.

Understandably, much focus has centred on safety and ethics, essential as AI continues to embed itself. And, with businesses finding AI ever more impactful, equal attention needs to be given to protocols for creating new products and services using AI.

But can traditional legal doctrines sufficiently safeguard IP in the product development process and through AI-generated innovations?

A history of court cases demonstrates the power and importance of IP – both its creation and protection. This, along with regulation, appears to be holding firm in the face of digital transformation. Nevertheless, the law needs to keep pace with technological advancement to ensure a fair playing field.

Recent examples include the launch of DeepSeek’s R1 reasoning model, a readily available large language alternative to ChatGPT. Whilst encouraging competition, concerns have been raised by ChatGPT about the uniqueness of the model, with suggestions that IP infringement may come into play.

Similar concerns have been raised in the creative world where, in a survey from insurer Hiscox in September 2024, titled ‘Art and AI’, 37% of art collectors raised concerns related to copyright infringement and AI generated art.

IP law

The threat of being copied, or having works plagiarised, is always a risk in business and this is present whether working collaboratively or from behind closed doors.

As it stands, IP law is complex but generally applies to works, designs etc created by humans.

Each set of circumstances, or facts, are examined on a case-by-case basis; however, it is important to be alive to copyright, database rights, patents, and trademarks when looking at AI systems and the outputs they generate.

Looking at this practically, and taking an example of English law, the following high-level ‘rules’ apply (currently):

1. Copyright

Section 9, Copyright Designs and Patents Act 1988 generally provides that the first copyright owner will be the author, or for computer generated works, the person by whom the necessary arrangements for the creation of the work are undertaken.

2. Application to AI systems

Training an AI system may lead to the use or copying of artistic or copyrighted works. Alternatively, a computer using AI could potentially create copyright-protected works. Protecting AI-created works with copyright involves complex issues, including meeting the originality test of the “author's own intellectual creation” ^[1] and determining ownership of the copyright in AI-generated works.

3. Infringement

Copyright may be infringed when training or using an AI system, e.g. if the process relies upon or copies protected copyright works without a requisite licence from the relevant copyright owner of that original work. Since the AI system is not a natural or legal person, where AI causes or creates IP infringement, potential liability for that sits with those deemed to be responsible for its actions^[2].

4. Wider regulation

When seeking to prove IP infringement in different countries and jurisdictions, legal tests and requirements likewise differ. Indeed, the legal and regulatory landscape in this area is still very much evolving.

How might things change?

We might expect to see key changes as part of policy reviews including the relationship between digital outputs which have relied on the creations of others.

Photo by Joshua Sortino / Unsplash

When developing AI, language models are taught (be that by humans or other technology). If their ‘curriculum’ of learning relies on data or works created by others, to which IP rights attach, this creates significant risk of a breach or claim.

A further challenge – particularly from the perspective of digital innovators - will be whether data (not typically generated by humans) attracts IP protections. If not, new regulation may be required to protect AI/digital outputs.

We may therefore see those seeking to exploit AI generated outputs (possibly with little regard to the IP of others), themselves seeking new laws to protect their own AI generated works. With various vested interests in play this may yet see a fair playing field emerge for the long term and across industry.

In the meantime, creative industries have been consulted on Copyright and Artificial Intelligence by the UK Government. Seeking to drive growth and innovation, and influence AI regulation worldwide, it will be interesting to see how the results influence change more broadly.

Continuing implementation of the EU AI Act and associated guidance and regulation, together with further global initiatives, will see ongoing developments in the mid-term. And in the UK developments include:

Legislation

• a potential move towards tighter regulation of the development, use and liability of AI systems^[3].
• proposals on AI regulation and connected purposes, addressing issues such as the creation of an AI Authority, transparency, accountability and safety^[4].

Public procurement

• new rules and guidance sources for procurement^[5]. The UK Government’s AI Playbook (findings of which are summarised in a Beale & Co article) also signals an increased focus on safe and responsible implementation of AI in public sector operations.

Contracts

• increased due diligence and negotiation of contract-specific terms and conditions for AI-related models or services may be required. Traditional forms of construction contract, for example, often do not cater for the use of AI tools or associated risks. In AI contracts for ownership, assignment or licensing, it is important to consider the position regarding AI-generated content or inventions.
• the potential application to smart contracts may also pose new legal and contractual challenges.

New precedents

• change may come in response to the outcome of Getty Images (US), Inc and others v Stability Al Ltd concerning a claim against an open-source generative AI company for copyright, database right and trademark infringement and passing off. We await the outcome as the liability trial is due to be heard this summer.
• we await the appeal hearing of the Court of Appeal’s earlier judgment in Comptroller - General of Patents, Designs and Trade Marks v Emotional Perception AI Limited, and clarity on patent protection in the context of and application to an artificial neural network (used in AI/machine learning). This is anticipated later this year^[6].

Data protection protocols and confidentiality

• further consideration of data protection related to AI or the risk of AI breaching confidentiality obligations.

Best practice checklist

In a world increasingly shaped by digital technologies, how should businesses address these issues? Steps include:

• Ensuring trustworthy AI and digital tools are used which are also accurate and dependable.
• Including an appropriate and proportionate standard of care or undertaking in your appointments and contracts.
• Carefully considering provisions relevant to IP ownership, and the use of AI, up and down the contractual chain and in line with the specific works/services to be performed. Contract terms should reflect the project and commercial-specific requirements.
• Having a clear specification or scope of work, and clear contractual provisions and risk allocation.
• Checking the terms of any licence for use of the AI system or software used since this may contain conflicting terms, exclude liability, or claim ownership of outputs generated.
• Clearly communicating the limitations of the AI system or output, not least to aid negotiation of terms/scope or manage expectations overall.
• Ensuring appropriate, timely training of employees, together with a clear understanding of codes of conduct or applicable policies and procedures.
• Supporting quality controls in place (including decision-making points triggering ‘human’ review and sign-off).

Photo by Amy Hirschi / Unsplash

Businesses should also implement robust and clearly visible procedures and carefully consider developing an AI and Data policy. This is important both from a liability and IP perspective. In the UK, for example, we are already seeing requests for evidence of this in tenders and contracts, with the approach encouraged by the UK Government’s AI Playbook.

Monitoring the effectiveness of AI policies and procedures and risk mitigation measures is equally important.

As with any design development, being alert to bias and the IP rights of others is key. Irrespective of design output, any breach of contract or IP can lead to significant costs, dilution of relationships or damage to business reputation. There may also be contractual indemnities linked to IP infringement or to confidentiality or data protection breaches.

Importantly, prioritise taking appropriate measures to protect your ideas, data, and IP early on. This could include having contracts in place and considering how such information is stored, labelled, shared and recorded.

Finally, always retain good records, including those aiming to demonstrate the agreed project arrangements and level of factual, technical and scientific knowledge, when developing and deploying any AI system.