Generative artificial intelligence is transforming how innovators research, develop and communicate ideas. Large Language Models (LLM’s) such as OpenAI’s GPT series, Google DeepMind’s Gemini, Anthropic’s Claude, and Meta’s LLaMA models can assist with brainstorming, drafting, research, summarisation and productivity, and are becoming increasingly common throughout research and development.
However, when confidential inventions, product designs, software developments or brand concepts are involved, AI must be used very carefully, if at all. Where AI is used, it is imperative that it is being used in an environment that protects confidential information and supports a robust intellectual property strategy.
Confidentiality Risks: Public AI Versus Closed AI Systems
The importance of keeping an invention confidential cannot be understated and is essential before filing any patent application. This is because most patent systems around the world will not grant patents for inventions which lack novelty, meaning an invention must not have been publicly disclosed before the relevant filing date of the patent application.
The provision of confidential information into LLMs therefore presents an inherent risk as there is currently no clear consensus as to whether the input of that confidential information into an open source LLM would constitute a public disclosure in all circumstances. Whilst the answer to this question will likely turn on the facts and the specific LLM in question, a key consideration in such cases is likely to be where and how the AI was used.
Consumer AI Platforms
Consumer-facing AI services may operate under standard user agreements and data policies that are not designed specifically for confidential innovation work.
Uploading prototype specifications, engineering drawings, source code, manufacturing processes, technical research, product development plans or unpublished brand concepts may create uncertainty regarding how that information is stored, processed or protected and ultimately whether it remains confidential after being used within such an AI platform.
The exact risk will likely depend on the provider, account type, contractual arrangements, and security settings. However, innovators should not assume that a consumer AI platform provides the same confidentiality protections as a professional adviser operating under formal obligations of confidence.
Closed and Controlled AI Systems
Where confidential intellectual property is involved, a closed or enterprise-controlled AI environment may provide a more appropriate solution.
Such systems typically incorporate stronger contractual protections, restricted user access, enhanced security controls, and governance arrangements designed to protect commercially sensitive information.
Even then, organisations should ensure that appropriate contractual safeguards, security measures, and internal policies are in place before using AI with confidential material.
The distinction is important. Consumer AI tools may be appropriate for general research, drafting assistance, and productivity tasks. Closed AI systems are generally better suited where unpublished inventions, confidential research, or commercially sensitive information are involved.
Regardless of the platform used, AI should support, not replace, informed human judgement. Decisions relating to patent filings, trade mark protection, and wider IP strategy should always be reviewed by appropriately qualified professionals.
The Limitations of Generative AI
As the use of generative AI becomes more widespread, businesses are increasingly using it to assist with drafting patent applications and other intellectual property documents. While AI can improve efficiency, careless use can put valuable inventions and future rights at risk.
The limitations of generative AI systems stems from the fact that they generate responses based on patterns in information rather than genuine technical understanding or legal judgment. While on the surface the results can often appear convincing, beneath the surface they can be littered with problems and thus should generally not be relied upon without appropriate review by a qualified attorney. Some examples of this are as follows.
Patent Drafting
Effective patent drafting requires far more than producing well-written technical text. A patent specification must describe an invention clearly enough that a skilled person could reproduce it, while also providing a sound legal basis for meaningful protection.
AI-generated drafts may contain technical inaccuracies, unsupported assumptions, fabricated references, or omit important aspects of an invention altogether.
A common misconception is that these issues can simply be corrected later. In practice, patent law generally prevents applicants from adding unsupported new technical matter after filing. If important technical information is missing from the original application, later amendments may not preserve the original filing date and could significantly reduce the value of the resulting patent.
Trade Mark Searches and Clearance
AI can be a useful brainstorming tool for developing potential brand names, but it cannot determine whether those names are legally available for use or registration.
A professional trade mark clearance exercise may involve assessing registered trade marks, pending applications, earlier commercial use, company names, domain names, reputation, and jurisdiction-specific legal considerations.
Launching a new brand based solely on AI-generated advice may result in infringement disputes, costly rebranding, and wasted marketing investment.
If you would like advice on trade mark searches and clearance, contact our trade mark attorneys at Secerna.
The False Economy of DIY AI Filings
Using AI to prepare patent applications or trade mark filings with the intention of asking an IP professional to "tidy them up later" can create a false economy.
What appears to be a saving at the drafting stage may quickly become more expensive if significant remedial work is required.
An IP professional reviewing an AI-assisted disclosure may need to identify the genuine inventive concept, separate valuable technical information from AI-generated content, correct inaccuracies, reconstruct missing details, and determine whether meaningful protection remains available.
The cheapest approach at the beginning is not always the most cost-effective approach over the lifetime of an intellectual property asset.
AI-Assisted Innovation Still Requires Human Inventors
AI is also raising important questions around inventorship and ownership.
Current patent systems generally require human inventors. As legal approaches to AI-assisted inventions continue to develop internationally, businesses using AI during research and development should therefore maintain clear records demonstrating the human contribution to an invention and the evolution of the final inventive concept.
Businesses should also retain records showing how AI tools were used during development, alongside evidence of the key human decisions that shaped the invention. Particularly, where generative AI is used, care must be taken in identifying any hallucinations produced by the AI that might embellish an innovator’s original idea. Whilst this may seem helpful, such hallucinations may in fact murky the water on who invented what and who owns the innovation and any intellectual property rights arising therefrom.
AI as Part of a Responsible IP Strategy
None of this means AI should be avoided. Used appropriately, AI can significantly improve efficiency throughout the innovation process. It can assist with many routine tasks, allowing innovators and IP professionals to focus on higher-value strategic work.
The important distinction is between using AI as a productivity tool with appropriate controls and human oversight, and using it as a substitute for professional intellectual property strategy and legal judgement.
Increasingly, innovative organisations and professional advisers will incorporate AI into their workflows. Those that benefit most are likely to be the organisations that combine technological capability with strong governance, appropriate confidentiality safeguards, and expert review.
Used in this way, AI complements professional expertise rather than replacing it.
"AI is becoming an essential tool for innovators, but the environment in which it is used matters. Closed, well-governed AI systems combined with expert oversight can support innovation, while uncontrolled use of consumer AI platforms may expose valuable intellectual property to unnecessary risk."
Jonathan Roberts, Senior Associate, Secerna LLP
Best Practice for Businesses
Consumer AI platforms should not be treated as repositories for confidential inventions, unpublished designs, technical specifications or proposed trade marks without first understanding the applicable confidentiality and data protection arrangements.
Many enterprise AI platforms also allow organisations to ensure that customer data is not used to train future AI models, supported by contractual commitments and enhanced administrative controls. Businesses should nevertheless review the specific terms, security features and governance arrangements offered by their chosen provider before uploading confidential information.
Where AI forms part of the innovation process, organisations should consider enterprise or closed AI environments supported by appropriate governance, contractual safeguards and expert human review.
Whilst Artificial Intelligence can accelerate innovation, protecting that innovation still depends on sound intellectual property strategy, careful control of confidentiality and experienced professional advice.
Industry guidance reflects these principles. The Chartered Institute of Patent Attorneys (CIPA) advises inventors and businesses to use AI tools responsibly, avoid entering confidential invention details into AI platforms unless appropriate confidentiality safeguards are in place, carefully verify AI-generated content, and seek professional advice where intellectual property rights may be affected. Organisations should also develop clear internal policies governing the use of AI throughout the innovation process.
Need advice on protecting innovation in the age of AI?
Whether you're developing AI-enabled technologies or incorporating generative AI into your research and development process, Secerna's Patent and Trade Mark Attorneys can help you protect confidential innovations, strengthen your intellectual property strategy and maximise the commercial value of your ideas.