“Getting a patent” is not the same as “writing a document that looks like a patent specification”.
“Getting a patent” involves understanding what is a possible invention, identifying what is new & inventive patent-eligible subject-matter, and then defining boundaries that will be effective in stopping competitors from commercially exploiting the invention. This involves developing a careful claim strategy. Done well, a patent becomes a powerful commercial tool. Done poorly, it’s a very expensive yet worthless piece of paper.
A real risk arises when you enter technical details into AI tools or search engines without any confidentiality protections. If there is no legal obligation on the tool provider to keep your input confidential, handing over details of your invention will simply result in you losing control over what happens to that information. Losing confidentiality and consequently novelty and you have lost an essential requirement for patent protection in countries like the UK and in Europe, where there is no grace period for self-disclosure. In simple terms: if all the details of your invention are made public before you have filed your patent application with a patent office, your chance of obtaining a potentially valuable patent for your invention in the UK or Europe will have disappeared in a few brushstrokes of a keyboard.
Using AI to help craft a patent specification may also generate another, often overlooked, risk. If AI tools are used to help create the invention itself, this could raise questions about who truly made the inventive contribution. Could the AI itself be considered an inventor? Different countries treat this differently. In some jurisdictions, only human inventors are recognised, and an invention created by AI may be a bar to patent protection. In others, if there is at least one human co-inventor patent protection can still be obtained. So, to the untrained eye, what looks like helpful technical input from an AI tool that you’d not thought of yourself can end up complicating inventorship, ownership and ultimately patentability.
Early-stage companies should ideally treat patents as part of their broader IP strategy involving trade secrets, design rights, copyright, and any other appropriate forms of IP. Strong IP rights are built on controlling the confidentiality of information about your proprietary technology, choosing what to disclose and when, and aligning protection with your commercial goals.
The takeaway is simple: before you paste details of what you want to claim is your invention into a public tool, regardless of whether it is a search engine like Bing or Google or an AI chat-bot like Claude or Chat-GPT, think carefully about what you are giving away.
If you would like to discuss any of the issues raised in this article, please either contact your usual Dehns attorney, or Dr Coreena Brinck, a UK and European patent attorney who is also Dehns’ Head of AI Development.