Artificial Intelligence

AI that needs a patent strategy to match

AI is not one technology, and protecting it is not one job. The same underlying methods appear in drug discovery pipelines, autonomous vehicles, telecoms networks and consumer products, and each context changes what can be protected and how. Protection also has to stand up before the European Patent Office, whose approach to AI and machine learning inventions turns on technical character and continues to evolve.

 

The strategic calls matter early. Which parts of an AI system should be patented, and which are better kept as trade secrets, particularly where model weights, training data or inference behaviour would be hard to detect in a competitor's product? How much should you disclose, and when, in a field moving this quickly? Who owns what when a system is built on open-source components, licensed foundation models or university collaborations? And where human inventors have worked alongside AI tools, how do you keep inventorship clean?

 

Our AI patent attorneys help clients make these calls deliberately rather than by default, and build portfolios that hold their value as the technology and the case law move.

 

A track record that started with the field

We have worked in AI patenting for over ten years, drafting applications for neural networks, hidden Markov models and long short-term memory architectures long before generative AI made the field a boardroom priority. In that time, we have helped global innovators shape their AI filing strategies from the outset, contributing to multinational portfolios that today rank among the largest in the world incorporating AI.

That early experience matters. We have seen how examiner practice on AI has developed at the EPO and UKIPO, and we know which drafting decisions hold up years later, through prosecution, opposition and enforcement. Our attorneys also speak at conferences and publish regularly on AI patentability.

 

How we build teams

An AI invention rarely lives in the abstract. It diagnoses disease, designs molecules, controls machinery or manages networks, and the strongest protection comes from attorneys who understand both the AI and the domain it serves.

 

Keltie's AI patent attorneys work alongside chemists, biologists, physicists and engineers across the firm, and we assemble each team around your technology. A team protecting an AI drug discovery platform looks different from one protecting a machine vision system or a network optimisation method, even though all three are "AI". We staff the team to fit the invention, rather than the other way around.

Our AI practice is co-led by Kimberly Baylis, who is recognised as a Rising Star by Managing IP in 2022, 2023, 2025 and 2026. 

 

Expertise across the sector

Our attorneys have technical backgrounds and prosecution experience across the AI landscape, from classical machine learning to the latest generative models, including:

 

• Machine learning and deep learning, including model architectures, training methods and inference
• Generative AI, large language models and foundation models
• Computer vision, imaging and signal processing
• Natural language processing and speech recognition
• Reinforcement learning and game-theoretic methods
• AI in embedded systems, control and edge devices
• AI hardware and accelerators

 

We protect AI inventions across the sectors where they are applied, including financial technology, telecoms, medical imaging and medical technology, drug discovery, consumer electronics, embedded control systems, automotive, wind energy and agriculture.

 

What we do

We support AI clients across the full IP lifecycle:

 

• Patent drafting, filing and prosecution before the EPO and UKIPO, with extensive experience optimising protection at the USPTO and other major offices
• Invention harvesting, helping research and engineering teams identify which innovations are worth protecting
• Freedom-to-operate analysis, clearance and due diligence searches in crowded, fast-moving AI landscapes
• Portfolio strategy, including the balance between patents and trade secrets for models, weights and training data, with detectability kept in mind from the first draft
• IP due diligence for fundraising, licensing and acquisitions, including assessing whether applications are likely to grant and whether they cover a company's key products, or those of its competitors
• Oppositions, appeals and enforcement advice, including at the EPO and the Unified Patent Court
• Commercial agreements: licensing, negotiations, collaborations, assignments and trade secret management
• Patent Box advice, particularly valuable for AI inventions where the AI supports a server-based or cloud-based service
• Client training on invention capture, AI patentability and trade secret handling
• Trade marks and design rights, where the commercial story extends beyond the patents
• Working with overseas law firms to craft specifications that stand the best chance before the EPO, often with advice before PCT filing

 

Start-up, scale-up and commercialisation

Many AI companies are in the start-up or scale-up phase. We are extremely familiar with the issues that arise at these stages, from first filings on a limited budget to preparing a portfolio for due diligence, and we customise our services and the timeline for protection to match your commercial plans.

If you are working with AI, we would love to hear from you.