AI is becoming more and more rooted in everyday life, making tedious tasks easier and more efficient. However, using AI for tasks which require rigorous diligence and involve confidential information, such as patent drafting, could be fatal. SMEs and inventors must carefully consider when AI is appropriate and when it will cause more harm than good.
The Chartered Institute of Patent Attorneys (CIPA) recently published their guidance on using AI tools as an SME or inventor. The guidance stresses the importance of being aware of the risks which come with using AI in patent work. AI undermines the two core pillars of the patent system: human invention and absolute confidentiality. AI errors in a poorly drafted application could cause problems for years to come, whilst the unintended disclosure of information, such as inputting your invention into non-confidential AI tools, could render your patent application wholly unviable.
The risks
CIPA underlines six key risks behind using AI tools for patent work:
- Confidentiality breaches: Once you input material into AI, you cannot guarantee what will happen to it. The information may be treated as disclosed to the public, meaning the invention can no longer be patented and trade secrets become public domain.
- Hidden inaccuracies: Errors produced by AI may not be easy to notice but could emerge during enforcement or litigation later down the line.
- Patentability failures: There are very strict requirements for patent applications which determine whether an invention meets the patentability criteria. For example, “sufficiency of disclosure” requires that the invention is described completely, allowing other skilled people to replicate it. AI may not describe an invention to the necessary extent.
- Human inventorship requirements: The named inventor(s) of a patent must be human. Depending too heavily on AI can overshadow human contributions.
- Evolving legal landscape: Regulations are constantly adapting to changing technology and applications filed now may be impacted by later legal developments, leading to issues.
- Data training risks: AI tools may use your input material as training data for future AI models which could benefit competitors.
What to do
It is vital to never input a description of an invention into AI tools before you have filed a patent application. AI tools with guaranteed data isolation and which promise confidentiality can be considered, but care should still be taken to thoroughly check the accuracy of the output and provide clear evidence of human contribution. You can monitor evolving AI regulations through ongoing guidance from the Chartered Institute of Patent Attorneys and the UK Intellectual Property Office.
It is recommended to seek counsel from a patent attorney who has knowledge of the technology and legal landscape, and can advise you on your use of AI. Patent attorneys are also vital sources of information regarding what constitutes a complete and viable patent application, and will be more attuned to AI errors.
At Mathys & Squire, we have patent attorneys who specialise in AI and have a deep understanding of the legal implications for both AI-based inventions and the use of AI tools. Please get in touch with a member of our team here if you are seeking advice.
To read the full guidance from CIPA click here.
