AI is fundamentally changing how companies invent, file, and protect their intellectual property. The efficiency gains are substantial, but so are the risks. Critical questions about who qualifies as an inventor, whether an AI-assisted invention is patentable, and what documentation companies need to protect their patent rights are no longer theoretical - they are here now.

Outlined below is what business needs to know, with particular focus on the Canadian legal context and offers practical steps to protect your competitive position.

Understanding AI's role in invention

The appeal of AI tools in the invention process is undeniable, and the benefits can be transformative. With skilled prompting, an AI tool can synthesise approaches from thousands of sources in seconds and apply them to complex technical problems. Many R&D teams are discovering that AI surfaces ideas they simply would not have considered on their own. When used as a collaborator in the invention process, AI can dramatically accelerate innovation. However, its use raises serious implications for patent rights when deployed as a substitute for human expertise in deriving the invention.

Using AI to create patentable inventions raises fundamental challenges regarding inventorship. When an AI tool plays a meaningful role in generating an inventive idea, determining who, or what, is an inventor becomes genuinely difficult. While inventorship requirements vary by jurisdiction, the patent right generally begins with the inventor. If the law recognises only humans as possible inventors, then AI cannot be considered one and protecting a purely AI-generated invention with a patent becomes impossible. If the definition expands to include AI models, such inventions may become patentable, though significant questions remain about how inventorship would be attributed and how it could be assigned and/or transferred to a company.

A set of patent cases, referred to as the DABUS cases, have attempted to name AI as an inventor on patent applications in various jurisdictions in a test of local patent laws. While the DABUS cases have not been universally rejected, most jurisdictions have done so.

The DABUS cases were rejected in the UK - including by the Supreme Court in 2023 - the US, Australia, Canada, and the European Patent Office. In Canada the DABUS application was rejected by the Patent Appeal Board (PAB) on the basis that an AI system is not a valid inventor. While rejected by the PAB, the administrative practice has not yet been tested in court, which leaves some legal uncertainty for Canadian filers.

For businesses, the takeaway is clear - if AI tools are part of your R&D process, build a robust paper trail to support human inventorship. Contemporaneous records showing how human team members conceived the inventive idea and directed the work, including any AI assistance, could be the difference between a valid patent and a successful invalidity challenge.

Patentability challenges

Using AI in developing inventions can also raise challenges related to obviousness. For a patent to be valid, the invention must be non-obvious to a person of ordinary skill in the art, having regard to the relevant prior art. An AI model is trained on vast amounts of prior art while arguably possessing little inventive ingenuity. As such, an AI model may effectively function as an ideal person of ordinary skill. Under this analysis, any solution developed primarily by the AI system could be deemed obvious.

To avoid potential obviousness challenges, the R&D process should position the AI system as a tool directed by human inventors. As with supporting human inventorship, maintaining a detailed paper trail documenting how humans guided the AI model is essential.

Currently, no definitive rules exist regarding what details in the paper trail will prove most valuable for overcoming inventorship and patentability challenges. Given this uncertainty, more documentation is better. At minimum, notes on how inventors used AI tools can help address potential issues during prosecution and subsequent validity challenges.

Prior art searches

Prior art searching is where AI currently delivers its most immediate and tangible value. AI-powered search tools can analyse patent databases, academic literature, and product documentation at a scale and speed no human team can match. For businesses, this capability provides earlier insight into existing art, enabling smarter patenting decisions about what to claim and where to file.  It is important to remember the duty of disclosure, particularly in the US, which requires submitting any relevant prior art.  As such, part of the searching process must also include tracking of search results for subsequent submission to the relevant patent offices.

But an important caution applies here. Uploading confidential invention details to a public or non-secure AI platform can create serious problems. Most consumer AI tools retain user inputs to train their models, and there is no guarantee that proprietary information will remain protected. Worse, disclosure to a non-confidential AI service could trigger novelty bars in jurisdictions with absolute novelty requirements potentially destroying the ability to obtain a valid patent. Businesses must ensure that any AI tools used in R&D are either hosted in a secure, confidential environment or covered by appropriate confidentiality and data handling agreements.

Competitive intelligence

AI tools provide businesses with a powerful advantage in competitive intelligence. Tracking competitor filing patterns, identifying freedom-to-operate risks, and spotting licensing opportunities once required extensive manual effort. AI makes all of this faster and more comprehensive, enabling companies to build patent strategies that respond dynamically to market developments and identify white space for new filings.

One area business often underestimates is the litigation exposure that accompanies AI-assisted patents. As these patents become more common, more frequent challenges can be expected that target whether inventorship was properly documented. A company that cannot demonstrate clear human involvement in the inventive process may find its patent difficult, or impossible, to defend. Documenting the process and providing a supporting paper trail at the outset is considerably less expensive than litigating it later.

Drafting considerations

Preparing and prosecuting patent applications can benefit from AI tools. While AI can be used by inventors to provide detailed documentation of the invention, its use in preparing patent applications should be closely monitored. Patent professionals consistently identify significant errors in AI-generated text in patent applications. These errors can include unsupported claim scope, inconsistent terminology, incorrect descriptions of parts of the invention, and statements that create prosecution history problems.

These are not minor editorial issues as they can directly affect the enforceability of the patent. When an application is filed, the disclosure required to support the claims cannot easily be amended. Accordingly, any issues with the description, for example hallucinatory AI statements, or incorrect technical descriptions may not be easily corrected during prosecution of the patent. Businesses should expect their advisers to treat AI output as a starting point only, applying rigorous review before anything is filed.

Practical steps for protecting AI-assisted inventions

  • Document your innovation process rigorously. Maintain contemporaneous records showing where human judgment shaped or overrode AI suggestions. If inventorship is ever challenged, this documentation will be the evidence that protects your patent.
  • Establish clear AI usage policies. Develop internal protocols that define how AI tools may be used in your R&D process, including requirements for human review, approval workflows, and documentation standards.
  • Work with trusted advisers. Experienced advisers who understand your technology and business will help ensure claims are built around the concrete technical improvements your invention delivers and that they meet patentability standards in each jurisdiction where you file. AI drafts require thorough examination by an experienced patent professional. The stakes with regard to claim scope, validity and enforceability are too high to bypass this step.
  • Stay current on legal developments. CIPO, the USPTO, and the EPO are all actively developing guidance on AI-related inventions. The rules are still being written and companies with active patent programs need advisers who monitor these developments closely.

The bottom line

AI is here to stay in the innovation process, and companies that master its use will gain significant competitive advantage. While the tools are changing rapidly, the legal fundamentals remain unchanged: inventorship still requires human contribution, patents still must clear rigorous legal hurdles, and the consequences of getting it wrong can be severe resulting in failed prosecutions, lost rights, and invalid patents. The opportunity is real, but so is the risk. Companies that move early to establish strong documentation practices and secure experienced legal guidance will be best positioned to capture the value of AI-driven innovation.

How Gowling WLG is helping

Gowling WLG has one of Canada's leading IP practices. Our patent team partners with businesses at every stage of the patent lifecycle from invention assessment and prior art searching through prosecution, portfolio management, freedom-to-operate analysis, and litigation.

As AI becomes integral to how companies innovate, we help clients establish the right IP foundations for documenting human contributions to AI-assisted inventions, structuring applications to meet Canadian and international patentability requirements, and building portfolios that withstand scrutiny.

Whether your business is beginning to integrate AI into its R&D process or you have an existing patent portfolio and want to understand your exposure, we are ready to help. Contact us to discuss how we can support your innovation strategy and help you build a patent portfolio that stands up to scrutiny.